CN101436179A - Method and apparatus for converting text - Google Patents
Method and apparatus for converting text Download PDFInfo
- Publication number
- CN101436179A CN101436179A CNA2007101244658A CN200710124465A CN101436179A CN 101436179 A CN101436179 A CN 101436179A CN A2007101244658 A CNA2007101244658 A CN A2007101244658A CN 200710124465 A CN200710124465 A CN 200710124465A CN 101436179 A CN101436179 A CN 101436179A
- Authority
- CN
- China
- Prior art keywords
- cabin
- sentence
- meaning
- storehouse
- text
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Machine Translation (AREA)
Abstract
The invention relates to language translation and provides a text conversion method aiming at the defects of unreasonable library building mode by adoption of words or sentences as units, unsmooth translated sentences and difficult realization of multilingual intertranslation in the prior art. The text conversion method comprises: receiving a first natural language text file, and calling a first natural language library to generate a smooth semanteme text file corresponding to the semanteme of the first natural language text file; and receiving a second natural language output command, calling a second natural language library, and generating and outputting a second natural language text file according to the smooth semanteme text file. The invention also provides a text conversion device. The text conversion method uses sentential forms and sentential cabins as the units to build the language library; generated sentences have smooth semanteme, and the language library has reasonable size; and the text conversion method can solve the problems that words fail to express the meaning when the words are used to build the library and the language library is enormous when the sentences are used to build the library, realize direct conversion from language to semanteme by conversing the natural language file into the smooth semanteme file, and realize intertranslation between any languages in virtue of conversion from semanteme to language.
Description
Technical field
The present invention relates to the language translation, more particularly, relate to a kind of text-converted method and device.
Background technology
The improvement of papermaking technology greatly facilitates the preservation of information, has promoted development of human society.The appearance of computing machine makes that the storage of information is more convenient, has accelerated the update of information storage means, and increasing industry has realized with no paper office.General text-processing technical method is that the character of various spoken and written languages is encoded at present, generates disk file then, and its typical case's representative is exactly the WORD of Microsoft.Although made things convenient for the editor of document, the disk file that this mode generates still can only supply people's reading and writing of languages separately, can't realize striding languages and read, and the foreigner who for example is ignorant of Chinese can't understand the document of being write as with Chinese.This makes intelligent highly developed today that translation still can only rely on manpower to finish.
In order to realize the automatic translation of document, people once did a lot of the trial.For example, because almost every kind of language is all on without exception the basis that is based upon combinations of words, so people have expected realizing automatic translation in the mode that makes up dictionary Chinese invention patent application CN200510018157 for example naturally; But the very fast discovery of people, often the language fails to express the meaning for the text that this mode obtains, and can't understand at all, and as if reason just is the complexity of the various and grammer of speech meaning, even if these two obstacles rely on current state-of-the-art technology also can't overcome.In this case, it is that unit makes up the sentence storehouse that people have turned to sight with the sentence, for example Chinese invention patent application CN03128953, CN200410021989, the mode of building the storehouse with sentence has solved the problem of building the aspect that has in the mode of storehouse that the language fails to express the meaning with speech, the sentence smoothness that translation obtains.But natural language is ever-changing, and various grammers vary, the head and shoulders above big imagination in sentence storehouse, and almost blast, under existing technical merit, the mode of this heaviness can't realize.This shows, lack the major reason that efficient and rational language library structure means are current automatic translation poor effect.
In addition, existing translation system is often only translated between two kinds of fixing language, for example, and between English and the Chinese, between Japanese and the Chinese, also do not have a cover can realize the system of multi-lingual intertranslation on the market.
In sum, need a kind of technology, can make up language library with reasonable manner, and the document that will use certain natural language to finish arbitrarily converts other Language Document to.
Summary of the invention
The technical problem to be solved in the present invention is, builds the storehouse the language fails to express the meaning, builds with sentence that the storehouse language library is huge etc. builds the irrational defective of storehouse mode with speech at prior art, and the defective that can't realize multi-lingual intertranslation, and a kind of text-converted method and device are provided.
The technical solution adopted for the present invention to solve the technical problems is: a kind of text-converted method comprises the steps:
S1, the reception first natural language text file, the meaning of calling the meaning of one's words of the corresponding described first natural language text file of first natural language storehouse generation is led to text;
S2, reception second natural language output command are called the second natural language storehouse, generate according to the logical text of described meaning and also export the second natural language text file.
In text-converted method of the present invention, described first language storehouse comprises the first sentence pattern storehouse and storehouse, first cabin, and described step S1 further comprises the steps:
S11, read in a sentence in the described first natural language text file;
S12, call the described first sentence pattern storehouse sentence that reads in is carried out sentence pattern coupling, output sentence pattern meaning of one's words code, and mark off the sentence cabin of the described sentence that reads in;
S13, call storehouse, described first cabin and cabin coupling is carried out in a sentence cabin that marks off among the step S12, an output sentence cabin meaning of one's words code;
S14, the described sentence pattern meaning of one's words code of foundation and described sentence cabin meaning of one's words code generate the meaning unifier sign indicating number of the corresponding described sentence that reads in;
S15, judge whether to arrive the described first natural language text end of file, if execution in step S16 if not, then reads in next sentence, gets back to step S12 then;
The meaning unifier sign indicating number of S16, all sentences that read in that will generate is stored as the logical text of described meaning.
In text-converted method of the present invention, the described first natural language storehouse comprises the first idiom storehouse, in described step S12, calling the described first sentence pattern storehouse carries out sentence pattern coupling to the sentence that reads in before, call the described first idiom storehouse and judge whether the described sentence that reads in is idiom, if then export idiom meaning of one's words code, meaning unifier sign indicating number as the described sentence that reads in forwards step S15 then to; If not, then call the described first sentence pattern storehouse sentence that reads in is carried out the sentence pattern coupling.
In text-converted method of the present invention, storehouse, described first cabin comprises the mould storehouse, first cabin and the first cabin eye bank, and described step S13 further comprises:
S131, read in a sentence cabin of the described sentence that reads in;
S132, call mould storehouse, described first cabin cabin mould coupling is carried out in a sentence cabin of reading in, if the match is successful, then export cabin mould meaning of one's words code, and mark off the cabin eye in described sentence cabin of reading in, forward step S133 then to; If it fails to match, then forward step S134 to;
S133, call the described first cabin eye bank cabin eye that marks off is carried out cabin eye coupling, output cabin eye meaning of one's words code, and generate the sentence cabin meaning of one's words code in corresponding described sentence cabin of reading in according to described cabin mould meaning of one's words code and described cabin eye meaning of one's words code, forward step S135 to;
S134, call the described first cabin eye bank cabin eye coupling is carried out in a sentence cabin of reading in, output cabin eye meaning of one's words code, the sentence cabin meaning of one's words code as described sentence cabin of reading in forwards step S135 to;
S135, judge whether to exist untreated sentence cabin, if, then read in next sentence cabin, get back to step S131 then, otherwise, forward step S14 to.
In text-converted method of the present invention, in described step S133 or S134,, then in the described first cabin eye bank, look unfamiliar into cabin eye meaning of one's words code for the unsuccessful cabin of coupling if eye coupling in described cabin is unsuccessful, export this cabin eye meaning of one's words code then.
In text-converted method of the present invention, described step S14 further comprises:
Signal selected in S141, reception first sentence pattern, select the sentence pattern meaning of one's words code of corresponding described first sentence pattern selection signal from the sentence pattern meaning of one's words code of output;
Signal is selected in S142, first cabin of reception, selects described first cabin of correspondence to select the sentence cabin meaning of one's words code of signal from the sentence cabin meaning of one's words code of output;
Sentence pattern meaning of one's words code that S143, foundation are selected and sentence cabin meaning of one's words code generate the meaning unifier sign indicating number of the corresponding described sentence that reads in.
In text-converted method of the present invention, described second language storehouse comprises the second sentence pattern storehouse and storehouse, second cabin, and described step S2 further comprises the steps:
S21, read in the meaning unifier sign indicating number of a corresponding sentence in the logical text of described meaning;
S22, call the described second sentence pattern storehouse, convert the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number to corresponding sentence pattern;
S23, call storehouse, described second cabin, convert the sentence cabin meaning of one's words code in the described meaning unifier sign indicating number to corresponding sentence cabin;
S24, according to sentence pattern that generates and sentence cabin generate and the corresponding described meaning unifier sign indicating number that reads in of output sentence;
S25, judge whether to arrive the end of the logical text of described meaning, if not, then read in the meaning unifier sign indicating number of corresponding next sentence, get back to step S22 then.
In text-converted method of the present invention, the described second natural language storehouse comprises the second idiom storehouse, in described step S22, before calling the described second sentence pattern storehouse, described method also comprises calls the described second idiom storehouse, judges whether described meaning unifier sign indicating number is idiom meaning of one's words code, if, then export the idiom of corresponding described idiom meaning of one's words code, forward step S25 then to; If not, then call the described second sentence pattern storehouse,, generate corresponding sentence pattern according to the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number.
In text-converted method of the present invention, storehouse, described second cabin comprises the mould storehouse, second cabin and the second cabin eye bank, described step S23 comprises that further calling described storehouse, second cabin and the described second cabin eye bank converts cabin mould meaning of one's words code and cabin eye meaning of one's words code in the meaning of one's words code of described sentence cabin to cabin mould and cabin eye, and looks unfamiliar into described sentence cabin according to described cabin mould and cabin.
In text-converted method of the present invention, described step S24 further comprises:
Signal selected in S241, reception second sentence pattern, select the sentence pattern of corresponding described second sentence pattern selection signal from the sentence pattern that generates;
Signal is selected in S242, second cabin of reception, selects described second cabin of correspondence to select the sentence cabin of signal from the sentence cabin that generates;
Sentence pattern that S243, foundation are selected and sentence cabin generate and export the sentence of corresponding described second language.
The present invention also provides a kind of text conversion device, comprising:
The logical text memory module of meaning;
The first natural language memory module is used to store the first natural language storehouse;
The second natural language memory module is used to store the second natural language storehouse;
The conversion process module links to each other with the second natural language memory module with the logical text memory module of described meaning, the first natural language memory module, comprising:
The first natural language processing module, be used to receive the first natural language text file, call the logical text of meaning of the meaning of one's words of the corresponding described first natural language text file of described first natural language storehouse generation, and the logical text of described meaning is stored in the logical text memory module of described meaning;
The second natural language processing module is used to receive second natural language output command, calls the logical text of described second natural language storehouse and described meaning, generates and export the second natural language text file.
In text conversion device of the present invention, the described first natural language storehouse comprises the first sentence pattern storehouse and storehouse, first cabin, and the described first natural language processing module comprises:
The first sentence pattern matching module is used for reading in the sentence of the described first natural language text file, calls the described first sentence pattern storehouse and carries out the sentence pattern coupling, an output sentence pattern meaning of one's words code and a sentence cabin;
First cabin matching module is used to receive the sentence cabin that the described first sentence pattern matching module is exported, and calls storehouse, described first cabin and carries out a cabin coupling, an output sentence cabin meaning of one's words code;
The logical text generation module of meaning is used to receive described sentence pattern meaning of one's words code and sentence cabin meaning of one's words code, generates the meaning unifier sign indicating number of the corresponding sentence that reads in, and generates the logical text of described meaning according to the meaning unifier sign indicating number of all sentences that read in that generate.
In text conversion device of the present invention, the described first natural language storehouse comprises the first idiom storehouse, and the described first sentence pattern matching module comprises:
The first idiom matching module is used to judge whether the described sentence that reads in is idiom, if, then export idiom meaning of one's words code, as the meaning unifier sign indicating number of the described sentence that reads in, if not, then call the described first sentence pattern storehouse and carry out the sentence pattern coupling.
In text conversion device of the present invention, storehouse, described first cabin comprises the mould storehouse, first cabin and the first cabin eye bank, and described first cabin matching module comprises:
The first cabin mould matching module is used to call mould storehouse, described first cabin cabin mould coupling is carried out in the sentence cabin of reading in, if the match is successful, then exports cabin mould meaning of one's words code and cabin eye; If it fails to match, then directly export described sentence cabin of reading in;
The first cabin eye matching module is used to receive cabin eye or the sentence cabin that the described first cabin mould matching module is exported, and calls the described first cabin eye bank and carries out cabin eye coupling, output cabin eye meaning of one's words code;
Sentence cabin code generation module is used to receive described cabin mould meaning of one's words code and cabin eye meaning of one's words code, generates the sentence cabin meaning of one's words code in corresponding sentence cabin of reading in.
In text conversion device of the present invention, the described first cabin eye matching module comprises:
New cabin eye processing module is used for when it fails to match, for the cabin eye distribution bay eye meaning of one's words code that it fails to match, preserving and export this cabin eye meaning of one's words code at the cabin eye.
In text conversion device of the present invention, the first natural language processing module comprises:
First selects signal receiving module, is used for reception and transmits first sentence pattern selection signal and first cabin selection signal;
The logical text generation module of described meaning receives described first sentence pattern and selects signal and first cabin to select signal, in sentence pattern meaning of one's words code of receiving and sentence cabin meaning of one's words code, select corresponding sentence pattern meaning of one's words code and sentence cabin meaning of one's words code, generate the meaning unifier sign indicating number of the corresponding sentence that reads in.
In text conversion device of the present invention, the described second natural language storehouse comprises the second sentence pattern storehouse and storehouse, second cabin, and the described second natural language processing module comprises:
The second sentence pattern matching module, be used for reading in the meaning unifier sign indicating number of the corresponding sentence of the logical text of described meaning, call the described second sentence pattern storehouse and convert the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number to sentence pattern, sentence pattern that output generates and described meaning unifier sign indicating number be remaining part after the sentence pattern conversion;
Second cabin matching module is used for receiving described meaning unifier sign indicating number through the remaining part in sentence pattern conversion back, calls sentence meaning of one's words code conversion the forming a complete sentence cabin, cabin that storehouse, described second cabin will be wherein, the cabin that output generates;
The natural language text file generating module is used to receive described sentence pattern and sentence cabin, generates and export the sentence of the corresponding described meaning unifier sign indicating number that reads in.
In text conversion device of the present invention, the described second natural language storehouse comprises the second idiom storehouse, and the described second sentence pattern matching module comprises:
The second idiom matching module is used to judge whether described meaning unifier sign indicating number is idiom meaning of one's words code, if then export the idiom of corresponding described meaning unifier sign indicating number; If not, then call the described second sentence pattern storehouse and convert the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number to sentence pattern.
In text conversion device of the present invention, storehouse, described second cabin comprises the mould storehouse, second cabin and the second cabin eye bank, and described second cabin matching module comprises:
The second cabin mould matching module is used to read in a sentence cabin meaning of one's words code, calls mould storehouse, described second cabin and converts cabin mould meaning of one's words code wherein to the cabin mould, exports described cabin mould and described sentence cabin meaning of one's words code remaining part after the conversion of cabin mould;
The second cabin eye matching module is used to receive the part that described sentence cabin meaning of one's words code is left through cabin mould conversion back, calls the described second cabin eye bank, converts wherein cabin eye meaning of one's words code to the cabin eye, and exports described cabin eye;
Sentence cabin generation module is used to receive described cabin mould and cabin eye, generates described sentence cabin.
In text conversion device of the present invention, the described second natural language processing module comprises:
Second selects signal receiving module, is used for reception and transmits second sentence pattern selection signal and second cabin selection signal;
Described natural language text file generating module receives described second sentence pattern and selects signal and second cabin to select signal, selects corresponding sentence pattern and sentence cabin in sentence pattern of receiving and sentence cabin, generates the sentence of the corresponding meaning unifier sign indicating number that reads in.
Beneficial effect of the present invention is, uses the language library that makes up as unit with sentence pattern and sentence cabin, and the sentence meaning of one's words smoothness of generation, and language library size reasonable can solve prior art and build the storehouse with speech the language fails to express the meaning, build the huge problem of storehouse language library with sentence; By the natural language document being converted to the logical document of meaning, can implementation language to the direct conversion of the meaning of one's words, thereby realize intertranslation between any language by the meaning of one's words to the conversion of language.
The objective of the invention is to overcome the above-mentioned shortcoming of prior art, provide a kind of user of only requiring to understand mother tongue, needn't learn again, can be common to the text-processing that multilingual people read and write interchange, thereby realize striding the method and apparatus of languages, accessible interchange.
Why background technology exists above-mentioned some shortcomings, and people have certain relation to the understanding between " human brain ", " spoken and written languages ", " computer " three itself and the three, the division of labor.To should be understood that between " human brain ", " spoken and written languages ", " computer " three itself, the three: computer can not be understood semanteme as human brain; The length of computer is storage and search.Human brain is just in time opposite with computer, can understand semanteme, but storage and search capability are all far away not as good as computer.Thereby human brain and computer have good complementarity.As for spoken and written languages, it is the product of whole human social development.When the mankind produced spoken and written languages, because the isolation of space-time, people were dispersed in many independently societies and live; Spoken and written languages also in these many communities, are slowly developing and development in the independent system separately.The written word is a symbol for recording human speech; Language is the instrument that the human thought exchanges.The natural language that spreads so far is all very perfect, and complete voice, grammer and vocabulary system are arranged separately.Just because of various spoken and written languages evolving development lentamente in independent system separately all.Between the different language literal, the voice difference, grammer is different, and the vocabulary character is varied especially.Background technology just by this difficulty have difficulty in taking a step.An important rule out in the cold is arranged here: various spoken and written languages are all formed vocabulary by character, and sentence formed in vocabulary, and article formed in sentence; Expressing the meaning of various spoken and written languages is multi-level, and word, speech, sentence, section, chapters and sections etc. are arranged, and be quite complicated.Its essential attribute is that sentence is the elementary cell of the The expressed meaning of one's words; The identical meaning of one's words can be expressed in the sentence of different language literal; Vocabulary is ambiguity and be univocality in sentence usually.So, can not exchange without barrier between the different language people, key is word between the different language literal, speech, the sentence not reciprocity also disunity of expressing the meaning.Equity and unified this key that we just catch word between the different language literal, speech, sentence to express the meaning: 1. utilize the discovery (seeing for details hereinafter) of my sentence pattern sense-group principle to simplify the complicacy of natural language, adapt to its dirigibility again; And dissolve the inconsistent difficult problem of grammer between them, grammatical phenomenon complicated, personalization is pulled on sentence pattern, even it is also extremely simple to allow a cabin comprise grammer.Sentence pattern cans be compared to the stem of the topic of filling a vacancy, and the sentence cabin is that topic is empty.2. grammatical analysis, semantic understanding are the things of human brain, are finished by human brain; Organize the expert according to sentence pattern, sentence cabin principle, sentence is carried out the meaning of one's words comparison arrangement of sentence pattern, two aspects of sense-group string, make word between the different language literal, speech, sentence express the meaning and be able to equity and unification, data such as the sentence pattern of putting in order out, sense-group string are built the storehouse respectively and are preserved.The content in these storehouses is exactly word between the different language literal, speech, an equity of expressing the meaning and unified.It is difficult that these need be understood also is that the thing of putting things right once and for all is finished by human brain.Yet 3. give computer regular, dull, loaded down with trivial details memory, search work.Sentence pattern sense-group model is dexterously with after computer technology combines; A concrete sentence, as long as it is inserted in the sentence pattern of coupling, its " sentence meaning=sentence pattern+sentence cabin "., allow computer need not to understand, analyze the simple operations of doing " sentence meaning=sentence pattern+sentence cabin ".So allow the human brain, computer can be complementary, and give full play to their strong point.4. make word between the different language literal, speech, sentence express the meaning and be able to equity and unification; Encode at the sentence pattern that so draws, sense-group string etc. again; In fact, just carried out multilingual unified meaning of one's words coding.Yet just can realize striding accessible interchange the in languages ground by operation to primitive coding and decoding.Such method does not change spoken and written languages itself; The user also just need not to learn again.
Yet, more specifically definite, the object of the present invention is to provide a kind of meaning of one's words comparison arrangement towards multi-lingual sentence pattern, two aspects of sense-group string; Then, being compared between that arrangement is come out, the different language literal word, speech, sentence towards these expresses the meaning and is able to equity and unified sentence pattern, sense-group string encode (meaning unifier sign indicating number); Utilize meaning unifier sign indicating number to generate the disk file of multi-lingual meaning of one's words intercommunication---the logical text of meaning; Allow the latter be common to the written languages communication method of striding that multi-lingual people utilize mother tongue read-write separately.
Description of drawings
The invention will be further described below in conjunction with drawings and Examples, in the accompanying drawing:
Fig. 1 is the schematic diagram of technical solution of the present invention;
Fig. 2 is the present invention is generated the logical text process of meaning by the natural language text file a process flow diagram;
Fig. 3 is the present invention is generated natural language text file process by the logical text of meaning a process flow diagram;
Fig. 4 is the structural representation of text conversion device of the present invention;
Fig. 5 is the structural representation of the first natural language processing module among Fig. 4;
Fig. 6 is the structural representation of the second natural language processing module among Fig. 4.
Embodiment
Below, in conjunction with particular flow sheet specific implementation process of the present invention is described.
Fig. 1 is the schematic diagram of technical solution of the present invention.The concept definition that will involved in the present inventionly have earlier a certain sense be explained as follows:
Sentence---in natural language, the elementary cell of the The expressed meaning of one's words is a sentence; The identical meaning of one's words can be expressed in the sentence of different language literal.Sentence can divide sentence pattern, sentence cabin two parts, and a sentence pattern comprises several cabins.
Sentence pattern---sentence pattern comes from the abstract of a class sentence, and is relatively stable in sentence, embodies the basic meaning of one's words of sentence and generic; Claim sentence pattern with such sentence structure frame part of formation.Sentence pattern embody the basic meaning of one's words of sentence, generic stride languages, towards the whole mankind; And its structural framing is towards concrete natural language.
The sentence cabin---the sentence cabin is to be embedded on this structural framing of sentence pattern, accepts the selection and the restriction of sentence pattern; Content in the sentence cabin can use the sense-group text strings to replace, and forms colourful, concrete sentence.These replaceable flexibly parts are the sentence cabin.Sentence cabin number, the sentence cabin meaning of one's words towards the whole mankind, stride languages; But precedence in the sentence pattern structural framing and interchangeable sense-group text strings are towards concrete natural language.
Sentence pattern sentence cabin illustration (# shows capable number):
As long as 1# is can (02828) 2 cabin of sentence pattern
2# is your as long as [1], and you will [2].1[is unremitting] the 2[success]
3# [1]and?you?will[2]. 1[Persevere]2[succeed]
4# Если[1]，то[2]. 1[вынастаиваете]2[будете?с?успехом]
Other languages of 5#
6# resembles (00892) 4 cabin of a sentence pattern
Is 7# [1] as [4] [3] [2]? 1[John] 4[Henry] the 3[effort] 2[work]
8# Does[1][2]as[3]as[4]？ 1[John]2[work]3[hard]4[Henry]
9# [1][2]так[3]как[4]？ 1[Работает]2[Джон]3[усерно]4[Генли]
Other languages of 10#
(00456) 3 cabin of the high sentence pattern of 11#
12# [2] [1] is higher than [3].2[wisdom] 1[value] the 3[ruby]
13# The[1]of[2]is?above[3].?1[price]2[wisdom]3[rubies]
14# [1][2]выше[3]. 1[Стоимость]2[мудрости]3[рубина]
Other languages of 15#
Last routine 1. 1#6#11# triplex row is respectively represented the basic meaning of one's words of expression and the generic of three sentence patterns, meets the needs of the world, strides the languages part; Wherein as " as long as can sentence pattern " the expression generic and the basic meaning of one's words, (02828) expression sentence model is a meaning unifier sign indicating number low word decimal number.
2. 2～5#, 7～10#, 12～15# represent the structural framing of three sentence patterns, towards concrete natural language.The left front part of every row is the framed structure of sentence pattern, is a cabin in the square bracket; Right back part is corresponding sentence cabin and content example.Wherein 2#7#12# is towards Chinese; 3#8#13# is towards English; 4#9#14# is towards Russian; 5#10#15# is towards other languages etc.
3. in the above-mentioned for example square bracket or the numeral of its front be a cabin number.The number (having 2 cabins, 6# that 4 cabins are arranged as 1#) in sentence cabin and the meaning of one's words that each cabin is represented meet the needs of the world, stride languages; And its precedence in the sentence pattern framed structure, in order to the sense-group string of filling towards languages (as [2] among the 7～9# precedence difference in Sino-British Russian sentence pattern; The sense-group string of filling is respectively: work, work, Д ж о н).
For grammer, the complex grammar phenomenon is all pulled on sentence pattern; It is also extremely simple that grammer is contained in the sentence cabin.Above-mentioned is exactly sentence pattern, the definition in sentence cabin and succinctly giving an example of explanation.
The sentence cabin is filled by sense-group commander's sense-group text strings or is formed (also can be interpreted as by the filling of speech string approx or form).But the size in sentence cabin has than big difference.Minimum sentence cabin only comprises a sense-group string; Maximum sentence cabin can comprise a subordinate clause or subordinate sentence.We are divided into two kinds in simple sentence cabin and complex sentence cabin to the sentence cabin:
Sense-group---sense-group is the equity and the unification of character, speech, phrase and the phrase it " meaning " of multiple natural language; It is the elementary cell of human thinking activities.Sense-group does not have the limit of languages, belongs to the whole mankind, strides languages; Also metabolism with human social development.
The corresponding expression in the multilingual literal of sense-group text strings---sense-group claims the sense-group text strings.Be called for short the sense-group string; It is the composition of filling the sentence cabin.The sense-group string divides single string, goes here and there two kinds again; The single string that is that only contains an original speech string; Form by two or more original speech strings, and be connected to multiple string with " ".
The simple sentence cabin---the sentence cabin that is no more than three sense-group text strings except that the function word of not expressing the meaning claims the simple sentence cabin.English as " aan the on to and " etc. do not ignore when not expressing the meaning; The measure word of Chinese is disregarded equally.Other languages are so analogized.Between the different language, this string that three strings are only required accordingly, the meaning of one's words is identical is just passable, does not require their succession unanimities.
The complex sentence cabin---greater than the simple sentence cabin, the sentence cabin of containing the cabin mould claims the complex sentence cabin.
Above give an example the sentence the sentence cabin all belong to the simple sentence cabin.Below this example sentence contain the complex sentence cabin:
1{the?fisherman}consents?to?return?2{the_feather_suit}，on?condition?that?3{fairy?dance?andplay?heavenly?music?for?him}.
Be that 3{ fairy maiden he dances and plays melody in the sky } condition under, 1{ fisherman } promise to give back the 2{ garment or robe made of feathers.
This example sentence cabin 1, a sentence cabin 2 also are the simple sentence cabins, only contain a sense-group string (English the former is single string, and the latter is multiple string).The cabin mould is contained greater than the simple sentence cabin in sentence cabin 3, belongs to the complex sentence cabin:
Cabin mould and cabin eye-utilize above-mentioned sentence pattern theory to analyze a sentence cabin content; The framed structure that draws as sentence pattern partly claims the cabin mould; Be embedded in the structural replaceable part of cabin mold framework and claim the cabin eye.Sentence cabin and cabin eye are the upper and lower notions; But the equal and opposite in direction of simple sentence cabin and cabin eye is to be not more than three sense-group text strings except that the function word of not expressing the meaning equally.
As above example sentence cabin 3{fairy dance and play heavenly music for him}3{ fairy maiden dances for him and plays melody in the sky } utilize the theoretical analysis of sentence pattern, can get the deliver from vault mould:
(00205)[1]+[2]+and+[3]+for?him
[1]+for he+[2]+also+[3]
(00205) is the cabin pattern number in this example; This cabin mould comprises 3 cabin eyes, and the content of three cabin eyes all is not more than three sense-group strings:
3{1[fairy]2[dance]and?3[play?heavenly?music]for?him}
3{1[fairy maiden] be that he 2[dances] and 3[play melody in the sky]
The sentence that sentence pattern, sentence cabin told in little idiom-too briefly be not enough to is little idiom.As: " How do you do? you are good! Get away! Get away! " or the like.
Meaning of one's words code, meaning unifier sign indicating number-divide another name so-and-so meaning of one's words code towards the meaning of one's words coding of sentence pattern, sentence cabin, cabin mould, little idiom, sense-group string etc.; Towards them and the meaning of one's words after hinting obliquely at the corresponding composition of other languages coding claim meaning unifier sign indicating number.
The logical text of meaning-by meaning unifier sign indicating number generates and can embody multi-lingual meaning of one's words intercommunication, and the disk file that can the carry out multilingual text-converted logical text that claims to anticipate.
The sentence cabin is filled by the sense-group text strings, forms colourful sentence under the tissue of sentence pattern.The sense-group text strings constitutes the least unit of sentence, and any cabin all is made up of the sense-group text strings.Fill a cabin by the sense-group text strings, form the simple sentence cabin; A cabin is filled in the combination of sense-group text strings and cabin mould, forms the complex sentence cabin.The meaning of sense-group text strings is sense-group, and sense-group is the least unit in the sentence implication.
Design philosophy of the present invention is aspect form, to make up a cabin with the sense-group text strings, in case of necessity in sentence cabin use sense-group text strings and cabin mould structure complex sentence cabin; Organize a cabin by sentence pattern, generate complete sentence.Aspect expressing the meaning, constitute a base unit of the cabin meaning of one's words with sense-group, be used in combination sense-group in case of necessity and the cabin mould meaning of one's words constitutes a cabin meaning of one's words; Organize a cabin meaning of one's words by the sentence pattern meaning of one's words, generate complete sentence meaning.When generating meaning unifier sign indicating number by natural language, at first determine the sentence pattern and the sentence pattern meaning of one's words, mark off a cabin immediately, the sense-group text strings in the distich cabin is mated with the cabin mould that may exist then, generates the corresponding sense-group and the cabin mould meaning of one's words, thereby constitutes the sentence cabin meaning of one's words.Make up by the sentence pattern meaning of one's words and the sentence cabin meaning of one's words at last, generate the sentence meaning.
Content in Fig. 1 technical solution of the present invention schematic diagram, frame of broken lines all has corresponding storehouse (or the data in the storehouse), and has the part towards each natural language in the storehouse; The part that meets the needs of the world, strides languages is also arranged; And can hint obliquely at mutually at multilingual.The associative operation that also contains them simultaneously.Being natural language below the frame of broken lines, referring to contain the natural language of corresponding literal, also is spoken and written languages mentioned above.Wherein first natural language, second natural language can be choosing wantonly in these natural languages.Above the frame of broken lines is to stride languages, and the meaning that meets the needs of the world is led to text.
In the frame of broken lines related, towards the storehouse of various natural languages, as Chinese language storehouse, english language storehouse, Russian language library etc.Each natural language storehouse comprises sentence pattern storehouse, sentence storehouse, cabin and idiom storehouse at least.Wherein, the sentence pattern stock contains the sentence pattern and the corresponding sentence pattern meaning of one's words thereof, and sentence storehouse, cabin further comprises mould storehouse, cabin and cabin eye bank again, its middle deck mould library storage cabin mould and the corresponding cabin mould meaning of one's words thereof, cabin eye bank storage sense-group text strings and corresponding sense-group thereof.Corresponding tables of the idiom library storage idiom and the idiom meaning of one's words or the like.
Be not difficult to find out from schematic diagram,, just can generate the logical text of corresponding meaning if 1. begin flow process upwards from certain natural language.2. downward from the logical text flow process of anticipating, just the natural language of the logical any appointment of text generation of meaning can be exported.If 3. flow process from bottom to top and from top to down, when generating the logical text of meaning, again by meaning unifier sign indicating number, the first natural language storehouse hinting obliquely at direct generation and exporting second natural language text to the second natural language storehouse etc.Suppose that the second natural natural language is that mother tongue, first natural language of certain customer group is certain foreign language that this customer group needs.Like this, people can utilize mother tongue to read foreign language; And common people shared with regard to becoming the logical text of meaning as long as a people read.
Fig. 2 is the present invention is generated the logical text process of meaning by the natural language text file a process flow diagram.As shown in Figure 2, at first,, read in a sentence in the natural language text file in step 200.Can wait and divide sentence according to the specific punctuation mark in the natural language such as comma, fullstop, branch.Next,, call the idiom storehouse of this natural language correspondence and judge whether the sentence that reads in is idiom in step 202, if, then directly forward step 226 to, generate the idiom meaning of one's words code of this idiom, and with its meaning unifier sign indicating number as this sentence; If not, then forward step 204 to, call the sentence pattern storehouse of this natural language correspondence the sentence that reads in is carried out the sentence pattern coupling, output sentence pattern meaning of one's words code, and mark off the sentence cabin of reading in sentence.Subsequently, in step 206, read in a ready-portioned sentence cabin, then in step 208, call the mould storehouse, cabin of this natural language correspondence cabin mould coupling is carried out in the sentence cabin of reading in, if the match is successful, illustrate that then this cabin is a complex sentence cabin, need further to divide, at this moment, carry out step 210, export cabin mould meaning of one's words code, and further mark off the cabin eye in this complex sentence cabin.It fails to match as if the cabin mould, and then described sentence cabin of reading in is a simple sentence cabin, and the sentence cabin of promptly reading in is a cabin eye, at this moment will directly carry out step 212, carries out cabin eye coupling.Subsequently, in step 214, the cabin eye bank of calling this natural language correspondence carries out cabin eye coupling to dividing the cabin eye that comes out, if the match is successful, and then in step 218, the output sense-group; If it fails to match, then illustrating does not still have this cabin eye in the eye bank of cabin, and promptly Dui Ying word may be the word that neologisms or user create certainly.At this moment,, look unfamiliar into sense-group, forward step 218 then to, the cabin eye meaning of one's words code that output generates for the cabin that it fails to match in step 216.Next,, use the cabin mould meaning of one's words code of output and sense-group to generate sentence cabin meaning of one's words code, judge whether also to exist untreated sentence cabin in step 222 then in step 220, if, then get back to step 206, if not, then carry out step 224.Because the sentence pattern of natural language and the sentence cabin meaning of one's words are various, its combination is ever-changing, therefore need the user between sentence pattern meaning of one's words code of exporting and sentence cabin meaning of one's words code, to select sometimes, promptly in step 224, receiving first sentence pattern selects signal and first cabin to select signal, sentence pattern meaning of one's words code and the sentence cabin meaning of one's words code of selecting the user to want then in step 226, use sentence pattern meaning of one's words code and the sentence cabin meaning of one's words code selected to generate the corresponding meaning unifier sign indicating number that reads in sentence.In step 228, judge whether to arrive the end of natural language documents, if, then forwarding step 230 to, the institute that use to generate has a mind to the unifier sign indicating number and generates to anticipate and lead to text, if not, then gets back to step 200, reads in next sentence.
Below with concrete sentence said process is described.
For example, if in step 200, the sentence that reads in is " as long as unremitting, you will successfully ".Through the judgement of step 202, this non-idiom, therefore turn to step 204, use Chinese sentence pattern storehouse to carry out the sentence pattern coupling, obtain sentence pattern " as long as [1] you will [2].", and the sentence cabin " 1[is unremitting] " and " the 2[success] ".In step 206, at first read in a cabin " unremittingly ", because this cabin is the simple sentence cabin, promptly this cabin is a sense-group text strings, therefore execution in step 212 after step 208 is judged is carried out cabin eye coupling.Because " unremittingly " is a habitual speech, so the cabin eye coupling of step 214 will be succeeded.Forward step 218 output sense-group to; And then in step 220 with this sense-group as sentence cabin meaning of one's words code.Through the judgement of step 222, return step 206 and handle sentence cabin " success " then, ensuing processing procedure is similar to distich cabin " unremittingly ".After handling all sentence cabins i.e. " unremittingly " and " success ", forward step 224 to.Because this sentence is very simple, corresponding sentence pattern and sentence cabin implication are unique, therefore will directly pass through step 224, generate the meaning unifier sign indicating number of sentence meaning " need only unremittingly, you will successfully " in step 226.Otherwise, will list a plurality of sentence pattern meaning of one's words of sentence pattern correspondence and a plurality of the cabin meaning of one's words of sentence cabin correspondence at step 224 place and select for the user.Through above-mentioned steps, sentence " as long as unremitting, you will successfully " is finished.When read in be one piece of article the time, need judge whether that also in addition other sentences will be handled (step 228), if do not have, then use the meaning unifier sign indicating number of all sentences to generate the logical text of meaning.
Fig. 3 is the present invention is generated natural language text file process by the logical text of meaning a process flow diagram.As shown in Figure 3, in step 300, read in a meaning unifier sign indicating number of a corresponding sentence in the logical text of meaning.Next,, call corresponding idiom storehouse of specifying natural language, judge whether this meaning unifier sign indicating number corresponding sentence in this natural language is idiom in step 302, if, then directly forward step 314 to, generate sentence to the unifier sign indicating number of should anticipating.If not, then carry out step 304, convert the sentence pattern meaning of one's words code in this meaning unifier sign indicating number to corresponding sentence pattern, then in step 306, with the cabin mould meaning of one's words code conversion in the meaning unifier sign indicating number is corresponding cabin mould, if no cabin mould meaning of one's words code, then direct execution in step 308 is converted to corresponding cabin eye with the sense-group (i.e. the sense-group of this cabin eye correspondence) in the meaning unifier sign indicating number.Then in step 310, use the cabin mould that generates and cabin to look unfamiliar to become to insert the sense-group text strings among the cabin.Because the same meaning of one's words is the corresponding a plurality of synonyms of possibility in a kind of language, therefore need between sentence pattern of exporting and sentence cabin, select, promptly receive second sentence pattern and select signal and second cabin to select signal,, use sentence pattern and the sentence cabin selected to generate sentence then in step 314 in step 312.Next, carry out step 316, judge whether to express one's ideas and lead to the end of text, if not, then get back to step 300, read in next meaning unifier sign indicating number, if, then forwarding step 318 to, this flow process finishes.
Below with concrete sentence said process is described.
For example, if the meaning unifier sign indicating number that reads in step 300 is " as long as unremitting, you will successfully " (for ease of describing, directly use the Chinese implication of this meaning unifier sign indicating number correspondence to describe, under together) herein.If the output language of appointment is an English, then in step 302, call the idiom storehouse of English, judge whether this meaning unifier sign indicating number corresponding sentence in English is idiom, through judging, the sentence of this meaning unifier sign indicating number correspondence is non-idiom, then carry out step 304, with sentence pattern meaning of one's words code " as long as [1] you will [2]." convert to corresponding sentence pattern " [1] and you will[2]. ".Because " unremittingly " and " success " is the simple sentence cabin, therefore will skip then in step 306, direct execution in step 308, the cabin eye that sense-group " 1[is unremitting] " and " 2[success] " are converted to correspondence are sense-group text strings " 1[Persevere] " and " 2[succeed] ".Then in step 310, use the cabin mould that generates and the cabin cabin of forming a complete sentence of looking unfamiliar.If there is synonym, then need between sentence pattern of exporting and sentence cabin, select, promptly receive second sentence pattern and select signal and second cabin to select signal in step 312, then in step 314, sentence pattern that use is selected and sentence cabin generate sentence.Through said process, meaning unifier sign indicating number " as long as unremitting, you will successfully " converting to english sentence " Persevere and youwill succeed. ".Next, carry out step 316, judge whether to express one's ideas and lead to the end of text, if not, then get back to step 300, read in next meaning unifier sign indicating number, if, then forwarding step 318 to, this flow process finishes.
Fig. 4 is the structural representation of text conversion device of the present invention.As shown in Figure 4, text conversion device of the present invention comprises the first natural language memory module that is used to store the first natural language storehouse, be used to store the second natural language memory module in the second natural language storehouse, be used to store the logical text memory module of meaning of the logical text of meaning of generation, with be used at first natural language, the logical text of meaning, the conversion process module of changing between second natural language, wherein further comprise the second natural language processing module that the first natural language text file conversion is converted to the second natural language text file for the first natural language processing module of the logical text of meaning and the logical text of will anticipating.
Fig. 5 is the structural representation of the first natural language processing module among Fig. 4.As shown in Figure 5, select signal receiving module comprising the first idiom matching module, the first sentence pattern matching module, first cabin matching module, the logical text generation module of meaning, first, wherein first cabin matching module further comprises the first cabin mould matching module, the first cabin eye matching module, sentence cabin code generation module and new cabin eye processing module again.
Fig. 6 is the structural representation of the second natural language processing module among Fig. 4.As shown in Figure 6, select signal receiving module comprising the second idiom matching module, the second sentence pattern matching module, second cabin matching module, natural language documents generation module, second, wherein second cabin matching module further comprises the second cabin mould matching module, the second cabin eye matching module and sentence cabin generation module.
Beneficial effect of the present invention is, uses the language library that makes up as unit with sentence pattern and sentence cabin, the sentence meaning of one's words smoothness of generation, and language library size reasonable, and can solve prior art is that unit builds the unreasonable of storehouse mode with speech or sentence; By the natural language document being converted to the logical document of meaning, can implementation language to the direct conversion of the meaning of one's words, thereby realize intertranslation between any language by the meaning of one's words to the conversion of language.
Data are prepared part, on sense-group sentence pattern model based, according to the principle of sense-group sentence pattern.Grammatical analysis, the meaning of one's words are understood, and these computers are insurmountable also to be that the thing of putting things right once and for all is finished by human brain.Organize the expert according to sentence pattern, sentence cabin principle, sentence is carried out the meaning of one's words comparison arrangement of sentence pattern, two aspects of sense-group string, data such as the sentence pattern that comparison is put in order out, sense-group string are built the storehouse respectively and are preserved.The work of this part is that data are prepared, and is finished by 1～6 step hereinafter.Step for the preparation of the inventive method data.It is in order to solve word between the different language literal, speech, the sentence not reciprocity unified technical matters of expressing the meaning that data are prepared; Organize expert's (using a computer) to utilize towards multi-lingual many technological means of bilingual sample text being carried out sentence pattern, two aspect meaning of one's words comparisons of sense-group string arrangement; Express the meaning and be able to the technique effect of equity and information materials such as unified sentence pattern, sense-group string so as to producing word between the different language literal, speech, sentence.Data prepare successively to relate to " sentence pattern acquisition system ", " the whole speech system in sentence cabin ", " volume of compartment detection system " and " whole storehouse coded system ".Describe with the step coupling system below:
1. sample language material
Be used to import multi-lingual many input steps to bilingual sample text:
Input or collect some, enough big multi-lingual sample texts or many to bilingual sample text.These multi-lingual sample texts should be considered different fields, subject, style, and its quantity also should reasonable distribution.Then, with the chapter unit establishment code them; Marking literary composition behind the code plants; The common composing document name of these two parts.As " 0001 English, 0001 Chinese, 0001 Russia; 0002 Chinese; 0002 English; 0002 Russia or the like.Same article not its filename of identical text kind version with the code associated; As " 0001 English, 0001 Chinese ", " 0002 English, 0002 Chinese " is respectively two of same article literary compositions and plants versions, as a bilingual sample to (English is the A language, and Chinese is the B language).Or the like by that analogy.
2. sentence pattern aspect meaning of one's words comparison arrangement
Organize the expert to multi-lingual many steps of bilingual sample text being carried out the comparison arrangement of the sentence pattern aspect meaning of one's words:
Organize the expert to utilize " sentence pattern acquisition system " to carry out the comparison housekeeping of sentence pattern aspect.The sentence pattern acquisition system, hereinafter to be referred as system, with multi-lingual sample branch storehouse for reading in.At first 1., read the sample sentence, read in the bilingual sample example sentence of A, B of the identical meaning of one's words at every turn.System is operated by the expert, whenever finishes an operation item and all waits for after the expert clicks and just continue operation, so that expert's approval or intervene.If the sentence cutting that the expert is read in system has objection, can intervene by record is flexible up and down and revise.2. declare little idiom then, if the expert assert that this sentence is to being little idiom then deposit little idiom storehouse (containing fields such as idiom sign indicating number, middle habit, English habit, Russia's habit) in; Otherwise 3. system joins sentence pattern automatically, joins operation if find the type example that the sentence pattern that is complementary jumps to hereinafter; The sentence pattern that does not have coupling.4. dig a cabin by what the expert intervened the new work sentence pattern, dig before the cabin expert and must fully understand sentence, be known what's what in whole sentence pattern, sentence cabin and start again after (detailed main points of hereinafter digging a cabin) etc. take into full account.The generic of earlier selected this new sentence pattern.Then the sample example on cut out A first cabin of speaking, then cut out speak first cabin of the corresponding meaning of one's words of B.System detects correct, is intervened cutting out second again by the expert; The 3rd ... until having dug all cabins of this sentence.A, B language sample is cut out a cabin just becomes the sentence pattern of A, B language later on.Carry out 5. the type example then and join, promptly system recovers this new sentence pattern to this distich example just now again.Assert through the expert and 6. related data to be preserved after correct, promptly this sentence pattern, sentence model and generic are deposited in sentence pattern storehouse (containing fields such as sentence pattern generic, sentence pattern sign indicating number, medium-sized, English type, Russia's type); This sample example deposits corpus A (containing a model, A language example, B speak routine) in the form of type, routine combination again.The sentence pattern collection of this sample example finishes.Then jump to again 1., descend the sentence pattern acquisition operations of a pair of sample example sentence.
The main points of sentence pattern aspect comparison arrangement are how to cut out a cabin, determine sentence pattern.The operation expert should do following consideration earlier: 1. semantic consideration is the sentence pattern that divides into relatively-stationary, that can be used as framework from which is semantically considered; Which is an interchangeable cabin that divides into, cuts out.2. grammer is considered, considers from the grammer aspect, and grammatical phenomenon complicated, personalization is all pulled on the sentence pattern; Even a sentence cabin have grammatical phenomenon also should be very simply place side by side, modified relationship.In addition, if run into, when more grammatical phenomenon is arranged, consider that then extracting the cabin mould makes the complex sentence cabin from semantically consider will be as the sentence cabin but content is more, longer.3. part of speech is considered, the part of speech in sentence cabin is confined to noun, number, adjective, multiple string as far as possible, and a few cases is just considered other part of speech (as verb, adverbial word etc.).4. representatively consider that sentence pattern is the framed structure part of a class sentence, should increase representativeness as wide as possible, before it topic be give over to the sentence pattern part can not be very little, very little hinder detecting of sentence pattern.Should note pronoun simultaneously, it include in sentence pattern or the sentence cabin directly influence representativeness and practicality.The operation expert just cuts out a cabin through after the consideration of above-mentioned four aspects, and it is just right to accomplish.
3. sense-group string aspect meaning of one's words comparison arrangement
The step of organizing the expert that the data after sentence pattern aspect meaning of one's words comparison arrangement step are carried out sense-group string aspect meaning of one's words comparison arrangement again:
The comparison of sense-group string aspect continues after the comparison arrangement of sentence pattern aspect, utilizes " the whole speech system in sentence cabin " to carry out at the sentence cabin; Corpus A is its input, and this system is also operated by the expert.At first 1., it is right to read sentence, reads in the record of corpus A, promptly passes through the sentence pattern collection, right in conjunction with the sentence that pattern exists with the type example.2. system from left to right gets a cabin then, promptly takes out the leftmost sentence of sentence pattern cabin content (the sequence number reservation of sentence cabin is constant also ignores).Utilize sense-group string storehouse 3. to cut again and join mark,, utilize A language vocabulary cutting B language, A, B language has been matched the speech string that the meaning of a word equates marked promptly to current sentence cabin content.4. declare a cabin then, judge that promptly it still is the complex sentence cabin that current content belongs to the simple sentence cabin.It is continuous down if the simple sentence cabin skips to the sense-group alignment; If not work done in the manner of a certain author complex sentence cabin, try 5. to join the cabin mould, if do not find in the storehouse, the expert assert will extract new cabin mould, new cabin mould is made in expert's intervention.The type example is joined then, examines survey cabin eye with the form that cabin mould, cabin eye mix.The speech meaning of the A that simple sentence cabin or cabin eye have, B words and phrases string is equal fully; If not etc., the 6. sense-group alignment that continues, the sense-group alignment operation is promptly by augmenting the meaning of a word; Plus-minus sense-group string length; Methods such as the multiple string of group are alignd multi-lingual vocabulary on sense-group.Then they are included in current sentence cabin or work as the front deck eye, if the cabin eye then finishes the whole sense-groups alignment of the cabin eye on this cabin mould, finish in just current sentence cabin.7. include mark then in, they are made mark, include on the room in the current sentence of sentence pattern cabin.Whether the sentence cabin of judging current example sentence then all disposes, and is not to jump the operation of 2. carrying out next sentence cabin; If current sentence is finished, system carries out some format checkings, 8. preserve then, with above-mentioned through the sense-group registration process, the sense-group text strings of meaning of a word equity deposits sense-group string storehouse (containing fields such as sense-group sign indicating number, middle string, English string, Russia's string) in; New work cabin mould is stored in mould storehouse, cabin (containing fields such as cabin mould sign indicating number, middle mould, heroes and models, Russia's mould).Whole example sentence is deposited in corpus B with the form of type example combination; It is right that 1. jumping reads in next sample sentence more then, carries out the operation of a new round.The library structure of language material B and language material A is identical; Different is, and language material B preserved is that content A, the B language of the inside, sentence cabin is to go up equal, unified sense-group text strings in " meaning " through the comparison of sense-group string aspect; Another is not both the complex sentence cabin and also contains the cabin pattern number.
The main points of sense-group string aspect comparison arrangement are the sense-group alignment, and sense-group aligns towards the sentence cabin, comprises 1.. extend, augment the meaning of a word according to example; 2. do not change plus-minus sense-group string length under the prerequisite of original word, word, be convenient to splicing; 3. adhesive tape is attached the speech string; 4. the meaning of a word augmented in morphological change; 5. group is multiple goes here and there; With reference to the above definition and the explanation in simple sentence cabin, for the more sentence cabin of some contents, should they be become " remove some function words only≤three speech strings " simple sentence cabin by the sense-group alignment.Another kind of situation is 6. they to be extracted the cabin mould as the complex sentence cabin, and the cabin eye has been " remove some function words only≤three speech strings " too then.The selection that extraction cabin mould is still knitted compound word is: the framed structure of approximate sentence pattern is arranged, wherein have as the sentence cabin and can extract the cabin mould with the situation that other speech string replaces, handle with cabin mould, cabin eye.Can't extract the utilization group compound word of cabin mould handles.Purpose is that multi-lingual vocabulary is alignd on sense-group in a word.
4. volume of compartment detects
Sentence pattern, two aspect meaning of one's words comparisons of sense-group string arrangement gained data are built the step that respective detection was preserved and carried out in the storehouse respectively:
Above to two aspect meaning of one's words comparisons of sentence pattern, sense-group string arrangement gained data, addressed bilingual little idiom to, sentence pattern to, sense-group string to the cabin mould to depositing idiom storehouse, sentence pattern storehouse, cabin eye bank (sense-group string storehouse) and mould storehouse, cabin respectively in; And some other intermediate data has also deposited in the corresponding storehouse simultaneously.
5. the type example compares index
Utilize the type example to plant the new literary composition of the every adding of beginning from the 3rd literary composition and plant the step of carrying out new round sentence pattern, sense-group string aspect meaning of one's words comparison arrangement than control indexes:
Above mention, collect enough big language material sample, concrete yardstick how; Organize the expert to carry out two aspects comparisons of sentence pattern, sense-group string housekeeping and will what be done; When just can add new literary composition plants.Address these problems, see that mainly the type example is such as He Erding.We can say and organize the expert to compare the housekeeping incipient stage, almost is every increase pair of sample example, has also increased a sentence pattern simultaneously.The sentence pattern number is divided by example sentence number=1 ,=100%; Along with the increase of workload, the increase of sentence pattern is more and more slower, and it is more and more faster that increase counted in example sentence.Just can consider that when equaling 1% the language material sample is enough big, expert's comparison work can be come to an end, can consider to add new literary composition and plant.Liken to the type example and to be this process of control indexes.
6. whole storehouse coding
Work out the step of meaning unifier sign indicating number for the uniform data of building storehouse preservation and process detection:
Above address and build the storehouse preservation and, need information encoded to concentrate in four storehouses such as sentence pattern storehouse, cabin eye bank, mould storehouse, cabin and idiom storehouse through detecting.This step utilizes " whole storehouse coded system " to carry out at these four storehouses.These four results that the storehouse is a data preparation step are depositing word between the different language literal, speech, sentence and are expressing the meaning and be able to equity and data and information such as unified sentence pattern and sense-group text strings.Wherein, 1. the sentence pattern stock is put sentence pattern, contains literary compositions such as sentence pattern generic, sentence pattern sign indicating number, medium-sized, English type, Russia's type and plants the type-word section; At least comprise sentence pattern generic, sentence pattern sign indicating number, a certain literary composition kind sentence pattern field.The sentence pattern of its identical meaning of one's words of each record storage, wherein the content of sentence pattern generic, two fields of sentence pattern sign indicating number is to meet the needs of the world, and strides languages; Other is deposited each literary composition respectively as fields such as medium-sized, English type, Russia's types and plants corresponding sentence pattern structural framing, and they are towards various natural languages.The sentence pattern storehouse can be divided into some minutes storehouses as required, and the branch storehouse can only contain the sentence pattern sign indicating number, two fields of sentence pattern planted in a certain literary composition.2. the sense-group text strings is deposited by cabin eye bank (claiming the sense-group storehouse again), the sense-group text strings from the sentence cabin, also be in order to fill the content in sentence cabin.Cabin eye bank contains literary compositions such as sense-group sign indicating number, letter are released, middle string, English string, Russia's string and plants string field; At least comprise sense-group sign indicating number, certain literary composition kind string field.Wherein the sense-group code field is deposited the coding of sense-group, and its content meets the needs of the world, strides languages; Other certain string field is to deposit certain literary composition to plant the sense-group string, plants towards certain literary composition.A sense-group deposited in record of cabin eye bank, the succinct explanation of this sense-group, and the sense-group text strings of this sense-group in each languages.Cabin eye bank also can be divided into some minutes storehouses as required, divide the storehouse can a connotation group code, two fields of certain language string.Can also form the branch storehouse according to single string, multiple string or sense-group domain classification in addition, for different applications.3. mould stock in cabin is put the cabin mould, comprises languages mould fields such as cabin mould sign indicating number, middle mould, heroes and models, Russia's mould.The corresponding cabin of each languages mould of the identical meaning of one's words deposited in the record in mould storehouse, cabin.It also can be similar to, and the sentence pattern storehouse is the same to be divided into some minutes storehouses as required.4. the idiom stock is put little idiom, comprises the little idiom fields of language such as little idiom sign indicating number, middle habit, English habit, Russia's habit.The identical little idiom of the meaning of one's words and the form of expression in each languages deposited in the record in little idiom storehouse.The content in little idiom storehouse is from the sentence pattern acquisition operations process, and some is too brief not conveniently classifies little idiom as with sentence pattern, the method comparison arrangement of sentence cabin, with what the sentence pattern existed, deposits little idiom storehouse in.Utilize " whole storehouse coded system " that meaning unifier sign indicating number is unified 5. to work out in above-mentioned four storehouses after arrangement.Meaning unifier sign indicating number is nybble fixed length 16 carry system codes.Wherein a left side the 1st～2 byte is high-order, the coding of representative group or class; It is low level that the 3rd～4 byte is played on a left side, in order to each member in group or the class is encoded.Two sign indicating numbers of 0000H, FFFFH high-order, that low level is central are empty, are not equal to the efficient coding that is encoded to of 0000H and FFFFH.See hereinafter embodiment for details.
Above the step explanation of data preparation finishes.
Hereinafter 7～9 steps are the step that the inventive method practices.Practice is in order to solve people's read-write that text that prior art produces with character code can only plant for literary composition separately, the technical matters that exchanges; Utilization generates the technological means of the logical text of meaning of multi-lingual meaning of one's words intercommunication towards the meaning unifier sign indicating number of establishments such as sentence pattern, sense-group string; Can utilize separately mother tongue to carry out the logical text of reading and writing meaning to stride the technique effect that languages exchange in order to produce the people of identical text kind not.Practice successively and relate to " certain language generates the logical TEXT system of meaning ", " the logical text language of meaning read-out system " and " striding languages AC network system ".Describe with the step coupling system below:
7. utilize meaning unifier sign indicating number to generate the logical text of meaning
Be used to accept mother tongue sentence of the every input of user or after the traditional text file whenever reads in a mother tongue sentence, utilize meaning unifier sign indicating number that its conversion is generated the sentence pattern of the logical text of anticipating and generates meaning lead to text, i.e. the step of " certain language generates the logical TEXT system of meaning ":
The object of the invention is addressed and is utilized meaning unifier sign indicating number to generate peculiar same disk file, why " peculiar " is to utilize mother tongue separately read and write for multi-lingual user, and why " same " is that to stride the languages sentence be same for the not identical text kind sentence generation of transmitting the identical meaning of one's words.The people of identical text kind do not utilize mother tongue separately to write, and generate the logical text of meaning and realize by " certain language generates the logical TEXT system of meaning ".Certain language generates the logical TEXT system of meaning and provides the just user who relates in the background technology and the not people of identical text kind to certain pragmatic family.This system correspondingly is divided into some kinds of language versions towards the people of identical text kind not, for example towards Chinese be Chinese edition, i.e. " Chinese generates the logical TEXT system of meaning "; Towards Great Britain and America people be english version, " English generate meaning logical TEXT system " or the like.The process of the logical text generation of meaning is the process of a coding in fact, promptly source text is compiled into the process of meaning unifier sign indicating number.
The logical text generation system of meaning can offer the user and directly import the mother tongue sentence and generate the logical text of meaning; Also can convert the text that uses traditional copy editor's software to generate to meaning logical text.No matter input mother tongue sentence still is that text is changed all is to be that the unit carries out with the sentence.1. the present invention conceives the logical text generation system of meaning reads in sentence, reads in a sentence or accepts sentence of input.2. look into idiom, search the whether little idiom of this sentence,, simultaneously the little idiom sign indicating number of this record is write the logical text buffer area (following abbreviation writes file) of meaning if the little idiom sentence that takes out this article kind from the idiom storehouse shows output.If not; 3. join sentence pattern, utilize the sentence pattern storehouse, find out the sentence pattern that mates most with current sentence, provide the sentence pattern that the version literary composition is planted.4. overlap sentence pattern then, current sentence is inserted in this sentence pattern, become the combining form of type example.If run into the sequence sentence cabin in the middle of this, system provides by user intervention the sentence cabin that links to each other separately.After all having confirmed current sentential form to this person, machine both sides, this sentence pattern sign indicating number is write file.Then 5. get a cabin, in type, routine combining form, take out the sentence cabin content on the left side, form a sentence cabin sequence number, a cabin number is write file according to current sentence cabin sign.Judge whether simple sentence cabin, current sentence cabin then.Be that 7. jump off in the simple sentence cabin; Otherwise utilize mould storehouse, cabin 6. to join the cabin mould, select the preferably cabin mould of coupling, cabin mould sign indicating number is write file.7. the meaning of a word is determined then, promptly the speech string in current sentence cabin is from left to right determined its meaning of a word one by one, is determined by user intervention during ambiguity.The support in sense-group string storehouse determined to have in the meaning of a word.After the current meaning of a word is determined, from the storehouse, take out this sense-group sign indicating number, this sense-group sign indicating number is write file.Be 8. string especially when not having this speech if sense-group string storehouse is looked into, represent that it is user's a self-word creation, system accepts user's explanation and note, and this speech string and explanation are written on file header, and head adds new string mark and writes file.This process 7.～8. may need to repeat repeatedly; It still is the cabin eye that its content belongs to a cabin, and there is control in system, if the cabin eye, system also will give deliver from vault eye sequence number in real time.After current sentence cabin content writes out, jump onto the operation of 5. carrying out next sentence cabin again.The whole end of operations in sentence cabin of full sentence are jumped onto operation or the conversion of 1. carrying out next sentence again.When entire article conversion or write end, system generates the logical text of meaning with the content deposit of the logical text buffer area of above-mentioned meaning.The logical text generation system of meaning end of run.
The logical text of meaning is made up of meaning unifier sign indicating number.More particularly the logical text of meaning is made up of sentence pattern sign indicating number, sense-group sign indicating number, idiom sign indicating number, cabin mould sign indicating number, sentence cabin sequence number, cabin eye sequence number, new string mark and section transformation character etc.These sign indicating numbers can be represented the multilingual literal; The logical text of meaning of the logical text generation system of the meaning of identical text kind generation equally is not these sign indicating numbers, so so long as the identical logical text of the meaning that article generated of the meaning of one's words is the same.This is one of " identical text ".
8. the logical text of meaning reads
When the user reads the logical text of meaning, provide the i.e. step of " the logical text language of meaning is read " of user's mother tongue according to the meaning unifier sign indicating number of the logical text of meaning:
Reading by means of " the logical text language of meaning read-out system " of the logical text of meaning realizes that the logical text read-out system of meaning also provides to the user, also is the user that relates in the background technology and the people of identical text kind not.This system also correspondingly is divided into some kinds of language versions towards the people of identical text kind not.For example towards Chinese be Chinese edition, i.e. " meaning logical text Chinese read-out system "; Towards Great Britain and America people be english version, " meaning logical text English read-out system " or the like.
The logical text generation of read-around ratio meaning of the logical text of meaning wants much quick.It is exactly the inverse process that generates, and belongs to the process of a decoding.1. read the sentence operation, read in with anticipating the unifier sign indicating number one by one, represent that when the next one is sentence pattern sign indicating number or idiom sign indicating number current sentence runs through to the logical text of meaning.The sentence that reads in is placed on and reads in buffer area, 2. gets idiom then, if what read in only is an idiom sign indicating number, to the idiom storehouse, takes out the little idiom that field planted in this idiom sign indicating number respective version literary composition, shows output, repeats 1..If 3. the sentence pattern sign indicating number gets sentence pattern, take out the sentence pattern demonstration output that this yard respective version literary composition is planted to the sentence pattern storehouse, insert a buffer memory simultaneously.4. get a cabin, in the sentence buffer memory, from left to right get a cabin, promptly take out the label in current sentence cabin, the code content in current sentence cabin is all taken out to reading in buffer area according to this label then,, just 5. get the cabin mould, it is inserted the cabin buffer memory if contain cabin mould sign indicating number here.The back has only sense-group sign indicating number and new string mark, takes out the sense-group string that the respective version literary composition is planted respectively, and it is inserted current sentence cabin or cabin eye.So far, the professional is clear, needn't be meticulous.The logical text read-around ratio of meaning generates much quick, is that decode procedure needn't be intervened at the family because read.But there are 6. word order intervention and loss compensation to need user intervention.Above addressing simple sentence cabin and cabin eye should be≤3 sense-group text strings, and the order of these three strings is above being ignored, thereby may not meet the custom of reading language, needs the user intervention adjustment.7. loss compensation should be≤3 sense-group text strings that ignoring with some function words is prerequisite, Zhong Wen measure word for example, English article or the like.Taking notice of when logical text is read that these have been lost recompensed.6.～7. need user intervention; Also can not need user intervention, take system's automated manner.As allow system based on context the association of speech string table look-up and compensate automatically or the word order intervention.Can also be the combination of the two, can be full-automatic automatically, just in case can not allow user intervention automatically again.6.～7. this also have a situation about needn't intervene is to require reading speed, also can not intervene when minding as user reader.Thereby these functions are to make the user optional.
9. stride the languages AC network
Above-mentioned certain language is generated the logical text of meaning and Yi Tong text language reads and be merged into certain language version, form the step of " striding languages AC network system " by a plurality of certain languages version:
1. certain language version, as towards Chinese be Chinese edition, claim the Chinese meaning logical text; Towards Great Britain and America people be english version, claim the English meaning logical text; So analogize as the logical text of Russia's meaning of one's words, the logical text of day meaning of one's words or the like.Each languages version generates the logical TEXT system of meaning by certain language and Yi Tong text language read-out system is formed.The people that each languages version offers corresponding languages use.The people of corresponding languages are users of corresponding languages version.Several 1. certain language version, 2. the webserver, 3. stride languages and exchange service centre and form " striding languages AC network system " jointly.
10. isolated plant
Implement the isolated plant of the inventive method, comprise signal reception, signal emission, phonetic entry, the phonetic synthesis output block similar to prior art; The peculiar parts that comprise also that the logical text mother tongue of the meaning of implementing this method is read, mother tongue sentence, mother tongue generate the logical text of meaning.
Below the step prepared for the data of present embodiment of 1～6 joint:
1. sample language material
It is right as bilingual sample how successively to choose which literary composition kind in the middle of the crowd is multi-lingual, should consider the problem of three aspects.The one, the characteristics difference that literary composition is planted; The 2nd, population planted in literary composition; The 3rd, development in science and technology and influence power in the world.We at first choose English and two literary compositions of Chinese are planted, and represent alphabetic writing and ideograph separately; These two literary compositions are planted to belong to and are accounted for first of half six big literary compositions of world population plant; It also is the natural language that world wide has considerable influence power.Right with English, Chinese as first pair of bilingual sample.Bilingual sample is to containing A language, B language.After the third spoken and written languages began to add, in the middle of A, B two languages, one should be to put in order through comparison, and another is that initiate literary composition is planted.
After the sample language material is ready to.Ensuing work is to organize the expert to utilize " sentence pattern acquisition system " and " the whole speech system in sentence cabin ", and (all taking interactive means) carries out the meaning of one's words comparison arrangement of sentence pattern, two aspects of sense-group string to these multi-lingual sample language materials.
Be that example is launched explanation with Chinese, English below.Its filename of identical text kind version is not right with the bilingual sample of code associated to get same article mentioned above, " 0001 English, 0001 Chinese ", " 0002 English, 0002 Chinese " ... input as " sentence pattern acquisition system "; English is the A language, and Chinese is the B language.
2. the meaning of one's words of sentence pattern aspect comparison arrangement
The comparison arrangement of organizing the expert to carry out the sentence pattern aspect utilizes " sentence pattern acquisition system " to carry out.We speak English as A; Chinese is spoken as B.It is right to be ready to bilingual sample.The expert operates the sentence pattern acquisition system then.As shown in Figure 2, module 201 beginning is read bilingual sample to 202, reads in A, B language sample file is placed buffer area, make pauses in reading unpunctuated ancient writings automatically 203 according to the end of the sentence symbol (as English " .:! ? ", Chinese ".:! ? ") respectively the cutting statement place action pane.The expert intervenes 204 allows the expert read sentence and intervention trend, if the expert thinks that automatic punctuate is wrong, certain sentence that can will read in by the demonstration knob of 205 clicks 203 word for word (speech) prolongs or returns buffer area to shorten it; If the expert thinks that the sentence that reads in is to correctly.For example read in be " How do you do? you are good! " this is to sentence, because too brief, the sentence that can not use sentence pattern, two aspects comparisons of sense-group string arrangement to handle is right, meets little idiom definition, expert's intervention makes to be declared idiom 206 and deposits idiom storehouse 207 in as little idiom.That for example read in is " the fisherman consents to return the feather suit; on condition that fairy dance and playheavenly music for him. dances for him and plays under the condition of melody in the sky the fairy maiden, and the fisherman promises to give back the garment or robe made of feathers." sentence example to 208 they as the sentence example mark is handled.Join mark 209 supports yet cut automatically by means of sense-group string storehouse 211, automatically cut and join the mark operation, the feature of this operation be A language cause from a left side right side one by one word take off, look into sense-group string storehouse 211, and therefrom take out equivalent string of B language, if this equivalent string is made mark to finding in the middle of the B language with it at example sentence, compare otherwise from sense-group string storehouse 211, take out the B next equivalent string of speaking again, finish until full sentence.The expert understands and intervenes 210 and provide a mutual window to the expert then, and the expert can click to mend and look into the sentence example to any one speech string, and this example is right until understanding fully.This section operation be for the expert read the sentence standby.Then, module is joined sentence pattern 212 automatically through 213 supports that utilize sentence pattern storehouse 224, for this example to searching the coupling sentence pattern.The feature of joining sentence pattern be earlier A is spoken example sentence from left to right word for word or word take out, scan and list the sentence pattern that contains this word or word; Be these sentence patterns that the circle is divided into plurality of sections with the sentence cabin again, the sentence pattern that each section can find in example sentence successively is that middle choosing is found, otherwise for not finding.If find by jumping to 220 operations after 214 judgements; If do not find the coupling sentence pattern then enter 215～219 and intervene the module group operation of newly organized sentence pattern by the expert.Here sentence pattern chooses 215 by the selected generic that will newly organized sentence pattern of expert, perform the preparation of newly organized sentence pattern, dig a cabin then: the expert cuts out A first cabin of speaking by 216, and cutouts replaces with [1], cuts out first cabin of the identical meaning of one's words by 217 at the B language then; What cut out replaces with [1] equally.This example become " the[1] consents to return the feather suit; on condition that fairydance and play heavenly music for him. dances for him and plays under the condition of melody in the sky the fairy maiden, and [1] promises to give back the garment or robe made of feathers." if editor finishes not 218 and do not obtain the expert and make the right information of sentence fully, starts the operation that dig out in next round sentence cabins by 219, repeats 216～217; This is taken turns cutouts and replaces with [2], and next round replaces to [3] again, this example become " the[1] consents to return[2], on condition that[3]. under the condition of [3], [1] promises to give back [2]." click until the expert by that analogy and finish the sentence pattern editor; The remaining part that contains [N] promptly is a sentence pattern after these operations cut out a cabin.Editor finishes not 218 and judges that digging sentence pattern finishes, transition, example joined 220,220 sentence patterns that this is newly organized, or the sentence pattern that finds in the storehouse when being come by 214 redirects and the example sentence progressive type example of operating are joined, promptly right each of this example partly sat in the right seat and go on their sentence pattern, show output by 221 then, this example become " the 1[fisherman] consents to return 2[the feather suit]; oncondition that 3[fairy dance and play heavenly music for him]. be that 3[fairy maiden he dances and plays melody in the sky] condition under, 1[fisherman] promise to give back the 2[garment or robe made of feathers]." revise or determine that 222 give the chances of a modification of expert, treat that the expert determines after, preserve data, A language example, the B that the sentence pattern after this type example is joined, sentence example are deposited in corpus A223 respectively in conjunction with pattern speaks on the routine field; Newly organized sentence pattern " the[1] consents to return[2], on condition that[3]. under the condition of [3], [1] promises to give back [2]." deposited in medium-sized, the English type-word section in sentence pattern storehouse 224 respectively.225 do not receive the information that the expert will come off duty if come off duty, jump back to automatic punctuate 203 and carry out the right operation of next round sentence; Do some operations of winding up (for example the bilingual sample of buffer area not being deposited with the old file name write-back to running through part) if receive the information redirect 226 of coming off duty, finish operation.
The main points of sentence pattern aspect comparison arrangement are how to cut out a cabin, determine sentence pattern.Dig under the prerequisite that a principle that cabin requirement expert grasps is exactly the assurance operability and pursue representativeness.Operability promptly allows computer need not to understand, analyze the operation of doing " sentence justice=sentence pattern+sentence cabin ".Representativeness be the sentence example that contains of sentence pattern what, it is good more to contain the many more representativenesses of an example.The principle that requires the expert to grasp illustrates as follows again:
Multi-lingual to semanteme consideration principle
From multi-lingual to, be bilingual right consideration semantically at least.If the condition permission is to get language as much as possible to extract carrying out sentence pattern simultaneously naturally; Exactly because can not just require be at least bilingual to carry out (more kinds of languages to remain later sentence pattern promote carry out).As:
We?used?to?go?to?the?movies?about?once?a?week.
Usually we divide out greatly weekly and see film one time.
This is right, and English can as a sentence cabin, but be inserted with " once " to " go to the movies " in the middle of the Chinese corresponding " seeing a film ".As a sentence cabin, " once in a week " of Chinese separated by other speech again " once a week ".Both of these case is all not all right, must be bilingual to semantically consideration all can be.This sentence to can:
1[We]used?to?2[go?to?the?movies?about?once]a?week.
Usually 1[we] weekly 2[divide out greatly and see film one time].
Representative consideration principle
What, the size in sentence cabin directly influences the representativeness of sentence pattern.Our principle be guarantee to pursue under the prerequisite of operability representative.Be advisable with what about sentence cabin size, hereinafter explain orally.Here explain representational problem earlier:
How many are there in your family? how many members are there in your family?
If this sentence to only " your family " as cabin; Available " his family; John ' s family; Your class " or the like fill.But because " Howmany " is semantic relevant with " family ".Chinese is asked " several mouthfuls of people " for " family "; But should ask " how many students ", or " how many people " for " class ".So only just poor as sentence cabin representativeness " your family ".If " How many " and " family " are made two sentence cabins, not only can correlate mutually on the meaning of one's words; And representativeness has also increased.
Simple and unadornedly consider principle accurately
In view of the location of the logical text of meaning is " simple and unadorned transmit exactly semantic ", attend to one thing and lose sight of another and do not have when anti-when digging a cabin, can revise the magnificent sentence of translating and be simple and unadorned literal translation, dig sentence pattern again.As:
There?can?never?be?too?much?deception?in?war.
There can never be too much deception in war.
It is not only magnificent but also terse that this example is translated sentence " there can never be too much deception in war ", but the sentence pattern collection is difficult to operation.Chinese is changed into simple and unadorned literal translation, and " many again crafty plots were not yet in the war." remake and dig a cabin processing:
There?can?never?be?too?much1[deception]in?2[war].
The 2[war] in many again 1[crafty plots] be not yet.
For?all?their?great?size，the?elephants?moved?absolutely?noiselessly.
Although the body of elephant is huge, its sound of walking about does not have yet.
This " some sound does not have yet "; Be difficult to implement to sentence pattern or certain cabin; Changing it into the simple and unadorned literal translation problem of " walking about but quietly " just has been readily solved:
For?all?their1[great?size]，2[the?elephants]3[moved]absolutely?4[noiselessly].
Although 2[resembles] the 1[body huge], 2[it] 3[walks about] get up 4[quietly].
The principle that grammer is simplified and traditional
Consider that from the grammer aspect grammatical phenomenon complicated, personalization is all pulled on the sentence pattern; Make a cabin inner syntax very simple.The above-mentioned example is not difficult to understand this point.In concrete operations, can suitably increase a cabin number and reduce a cabin complexity, few big sentence cabin (knowing clearly hereinafter) of doing of trying one's best.
1[She]never?2[comes]but?1[she]3[brings?something?for?the?children].
1[she] do not have the 2[fork-like farm tool used in ancient China one time] be not to bring some things for children with regard to 3[].
As increase a cabin, [3] [4] complicacy is made in sentence cabin [3] just reduced.As:
1[She]never?2[comes]but?1[she]brings?3[something]for?the?4[children].
1[she] do not have the 2[fork-like farm tool used in ancient China one time] be not just to be children 4[] bring some things of 3[].
Again in the following example:
1[He]did?not?see?that?2[the?longer?he?spoke?the?less?his?people?believed?him].
1{ he do not find out, 2[he say for a long time more, his same clansman does not believe him more }.
Increase sentence cabin number, change [2] and add [4] [5] for [2] [3]:
1[He]did?not?see?that?2[the?longer]3[he?spoke]the?less?4[his?people]5[believed?him].
1[he] do not find out, he says 3[] must to get over 2[of a specified duration], his same clansman of 4[] 5[does not believe him more].
1[She]saw?playing?in?2[the?garden]a?group?of?children3[aged?from?six?to?ten].
1[she] see between the about 3[of a group six years old to ten years old] and child in the 2[garden] lining plays.
Sentence cabin [3] is increased to a cabin [4] [5]:
1[She]saw?playing?in?2[the?garden]a?group?of?3[children]aged?from4[six]to?5[ten].
1[she] see the about 4[six of a group] year is to 5[ten] year old between child 3[] in the 2[garden] lining plays.
The principle of part of speech and replaceability
The sentence cabin is the part that can be replaced by other vocabulary, and interchangeable vocabulary is many more, and substitutability is strong more; The representativeness of sentence pattern is strengthened.The part of speech of vocabulary is confined to number, noun, adjective, multiple string as far as possible in the sentence cabin, and a few cases is just considered other part of speech (as verb, adverbial word etc.).If will give the part of speech row who does a cabin priority ranking, that at first is exactly string, special-purpose string, inferior then name string, adjective string ..., consider the verb string at last.What least consider is preposition and conjunction.Just preposition, conjunction are nearly all received and are returned the sentence pattern part.
The sentence pattern speech should not be very little principle
It is very simple to require the sentence pattern collection to reach a cabin inner syntax, and complicated grammatical phenomenon is pulled to the greatest extent in sentence pattern; Cut out the sentence pattern that stays behind the cabin, contained sentence pattern speech should not be very little because very little detecting of inconvenient sentence pattern.
Optimal situation is that each cabin all has the sentence pattern speech before, does not just have the situation in sequence sentence cabin.Principle is any one sentence pattern, and the framed structure in any literary composition is planted must have one or more text strings as the sentence pattern speech.Do not allow multi-lingual central, certain literary composition is planted even is not had the sentence pattern speech, and Chinese is because the most succinct, and this situation happens occasionally, once take place and must do over again, must avoid.
Several cabins, i.e. sequence sentence cabin problem appear sometimes continuously.Be the size in a cabin in addition, few as far as possible big sentence cabin problem of doing, these all should not be relevant very little with the sentence pattern speech, and these situations are set forth under related heading.
The principle that many sequence sentences cabin is avoided as far as possible
1[I]2[get?to?work]at?3[nine?o＇clock]every?morning.
Nine o'clock of 3[every morning] 1[I] 2[starts working].
Go up routine English and occur [1] [2] two sentence cabins continuously; Corresponding Chinese then becomes [2] three sequence sentence cabins, [3] [1].The sentence cabin links to each other and claims the sequence sentence cabin more than two or two; The sentence cabin links to each other and claims many sequence sentences cabin more than three or three.The sequence sentence cabin not only has the too little fraud of sentence pattern speech, and the cover sentence pattern also must manual intervention.Particularly three or more situations should be avoided as far as possible continuously.As above example reduces to two sentence cabins does not just have having considered of this:
I1[getto?work]at?2[nine?o＇clock]every?morning.
Nine o'clock of 2[every morning] I 1[starts working].
Few as far as possible principle of doing big sentence cabin
Above address a cabin and vary, minimum sentence cabin only contains a sense-group text strings; Maximum sentence cabin can comprise a subordinate sentence or subordinate clause.We should try one's best to lack and do big sentence cabin when doing the sentence pattern collection.What so, how to grasp this and lack as far as possible, be yardstick with? be as the criterion with " can only do like this ".Sentence is right for example:
Can，you?guess?1[what?I?was?doing]2[this?morning]？
You can guess 2[this morning] 1[I what is being done]?
I?have?forgotten?1[what?time]he?said?he?2[had?dinner]3[last?night].
I forget him his 3[last night] be 1[when] supper had of 2[].
What?were?you?doing?when?1[I?called?you?on?the?telephone]？
I phone you 1[] time, what are you doing?
I?have?forgotten?1[what?he?said?his?address?was].
I forget 1[he say him the address where].
Above four sentences right, draw a subordinate sentence by what, when respectively.First pair can separately make two sentence cabins to subject-predicate, time adverbial.Second pair can make three sentence cabins.The 3rd pair can not be separated the adverbial modifier, can only make a sentence cabin; The 4th couple of What puts question at predicative, can not separate, and can only make a sentence cabin.Third and fourth two couple " can only do " like this, and Here it is lacks the yardstick that the big sentence of work cabin will be held.
Dig the authorization principle behind the cabin
Dig and must conscientiously authorize after the cabin: with sentence pattern, so that each cabin authorize respectively.See sentence pattern justice earlier, must not have adhesion with sentence cabin content all from the sentence pattern speech.Authorize each cabin then, they must be interchangeable, separate with sentence pattern, with sentence pattern justice any adhesion are not arranged.If certain cabin and certain sentence pattern speech adhesion to some extent, necessary modification.For example:
When?do?you?think?1[the?meeting?will?be?held]？
Do you think when the 1[meeting is held]?
Draw sentence pattern like this, the sentence cabin is not right, sentence cabin content " when " has adhesion with sentence pattern speech " When ".Should be revised as:
When?do?you?think1[the?meeting]will?be?2[held]？
You think the 1[meeting] can when 2[holds]?
3. the meaning of one's words of sense-group string aspect comparison arrangement
The comparison of sense-group string aspect continues after the comparison arrangement of sentence pattern aspect, utilizes " the whole speech system in sentence cabin " to carry out at the sentence cabin; Above-mentioned corpus A is its input.As shown in Figure 3: by module beginning, continuous 301 beginnings, this work next end last time that can continue; Also can just begin.Reading sentence then, to read sentence example to 302 from corpus A303 right, and the sentence pattern that these examples are right, a sentence cabin are marked.On continuous for example " the 1[fisherman] consents to return 2[the feather suit]; on conditionthat3[fairy dance and play heavenly music for him]. 3[fairy maiden is that he dances and plays melody in the sky] condition under, 1[fisherman] promise to give back the 2[garment or robe made of feathers].Yet " get a cabin 304, take off one of them cabin (with marked serial number order [1], [2], [3], two language example sentences carry out simultaneously, and their particular locations in sentence pattern are unordered).As take off 1 " 1[fisherman] 1[fisherman] "; Get 2 " 2[the feather suit] the 2[garment or robe made of feathers] " the rest may be inferred.Cut and join mark 305 and utilize the support in sense-group string storehouse 314 to cut to join the mark operation.The feature of this operation is that A language example is caused right side word cutting-out one by one from a left side, looks into sense-group string storehouse 314, and takes out equivalent string of B language from 314; If this equivalent string finds the speech string into the semanteme pairing in the middle of example sentence is spoken to B, it is made mark, otherwise unpaired, do not mark; So circulation finishes until full sentence.Yet declare a cabin, judge that this sentence cabin of operating is that simple cabin denys? if current that get is " 1[fisherman] 1[fisherman] ", declaring is simple cabin 306; By 307 redirect sense-groups alignment 313.If current that get is " 3[fairy dance and play heavenly music for him] 3[fairy maiden be that he dances and plays melody in the sky] ".Be that complicated cabin is then descending, looking into cabin mould 308 utilizes mould storehouse, cabin 309 to join the cabin modulo operation, it is characterized in that earlier with sentence cabin content from left to right word for word or word take out, scanning is also listed the cabin mould that contains this word or word, be these cabin moulds that the circle is divided into plurality of sections with the cabin eye again, the cabin mould that each section can find in the content of sentence cabin successively is that middle choosing is found, otherwise for not finding.Do not find not 310,, accept the operation (215～219 operation gathered in similar sentence pattern) that the expert makes new cabin mould, make new cabin mould as " [1] [2] and[3] for him[1] be his [2] also [3] " through making new mould 311 if do not have.And new work cabin mould is included in middle mould, heroes and models' field in mould storehouse, cabin 309 respectively.If do not find not the 310 cabin moulds that find coupling; Continuous cabin eye is handled 312 contents with its middle deck eye and is proposed, 313 pairs of whole sentence cabins of control sense-group alignment or the operation of cabin eye one by one, and they constitute a partial circulating; 313 towards current sentence cabin or cabin eye, accepts the expert at sense-group string storehouse 314 and intervenes the operation of carrying out the sense-group alignment.Through the sentence cabin behind sense-group alignment or the new work cabin mould or cabin eye be remove some function words only≤unit of three speech strings.
Behind the end of operations of 313 pairs of current sentences of sense-group alignment cabin (comprising the complex sentence cabin), fit into sentence pattern and make corresponding mark in the cabin after including mark 315 in and will put; Whether sentence is finished in the right sentence cabin of sentence entirely to finishing not 316 judgements, if do not finish, gets the operation that a next cabin is carried out in a cabin 304 through 317 redirects; If finish, preserve.Carry out format checking by 318 pairs of full sentences of format checking, deposit corpus B319 then in.Notice that the sentence cabin in the middle of the corpus A is square bracket []; Here they are changed into braces { }, because of having the complex sentence cabin, its middle deck eye square bracket [].This example be " the1{fisherman}consentsto return 2{the_feather_suit}; on condition that 3{1[fairy] 2[dance] and3[play heavenly music] forhim}. is 3{1[fairy maiden] be that he 2[dances] and 3[play melody in the sky] condition under, 1{ fisherman } promise to give back the 2{ garment or robe made of feathers." judge that at last the expert comes off duty not 320, if do not have, reading in sentence through 321 redirects, to read in next sentence example to 302 right, carries out aforesaid new round operation; If the decision of coming off duty is arranged, carry on the rounding off work by finishing 322, flow process finishes.
The operation of above-mentioned " sense-group alignment " is exactly that multi-lingual vocabulary is alignd on the meaning of one's words.The principle measure that requires the expert to grasp illustrates as follows again:
According to example, extend or augment the meaning of a word
Nothing can be 1[wholly beautiful] that is not 2[useful]. everyly use without 2[] just can not the 1[perfection].
Wherein the ＂ useful ＂ meaning of a word has " useful, useful, useful ", selects them all to be not suitable for; Augment " application " meaning of a word item.
1{She}was 2{strong}, for all 1{she}was so 3{small}.1{ she though the 3{ slight of stature, very 2{ is solid }.As: wherein small is little, augments " slight of stature " meaning of a word item.
I contract a serious illness I am very ill..Verb " am " is augmented the meaning of a word of " trouble ".
Learn the truth understands truth, and Learn augments " understanding " meaning of a word item.
Do not change under the prerequisite of original word, word and add and subtract string length, be convenient to splicing
I please you every other day teach me I ask you to teach me every other day..Wherein ＂ teach ＂ v has " lecture, teach " meaning of a word; Subtracting the speech string long is that " religion " is convenient to splicing, increases " religion " meaning of a word item.
Sttend school entrance; Sttend adds " going into " meaning of a word item, and " adding " released in letter; School adds " " meaning of a word item, and " school " released in letter.
Adhesive tape is attached the speech string
Become " benefit, good thing, good intention, handy " etc. (Chinese Fu Sui measure word, latter's otherwise processed) as " good is good " adhesive tape.
Become " word " etc. as " word speech " adhesive tape.
The meaning of a word (participle, the meaning of one's words that comparative degree etc. are expressed increases the corresponding meaning of a word) augmented in morphological change
Been augments " still, how " meaning of a word; Punished augments " punishment " meaning of a word;
Best augments " best " meaning of a word; Had augments " in the past " meaning of a word;
Done augments " carried out, finished " meaning of a word; Villages augments " a plurality of rural area " meaning of a word;
The group compound word
Original speech string is connected to compound word (multiple string) with "-".Their fundamental purpose is to be convenient to embody better the meaning of a word; So that the not alignment that vocabulary is expressed the meaning between the identical text kind.Mainly contain following several situation:
1. directly speech is translated in the time of can not expressing, and presses sense-group and forms multiple string:
You were gone literal translation should be " you walk (being) "; " " derive from past tense, so form multiple string " you have walked you_were_gone "
Visiting I knew_nothing_about_it of call_on_me knows nothing
Compelled_to_go simply must go
Show_himself_in_his_true_colours reveals original shape, reveals true qualities
2. succinctly to dress, complicated signifier forms multiple string
English " per-mu grain yield, every mu of grain yield "; In be " per mu yield ".Multiple string " per-mu_grain_yield " formed in English, in dress.And for example:
Late_at_night late into the night
Down_to_the_countryside goes to the countryside
Fight_it_out_to_the_end fights to the last ditch
3. have escape to represent a things, or the composition of regular collocation is gone here and there again
The sleeping knows_nothing of fell_asleep knows nothing
Live wire under the come_back_from_the_front (the front front, come comes, back returns)
Sweat profusely for the_sweat_was_pouring_down;
The pictures_it_have_just_taken recent photograph
Put_my_finger_on points out
4. the meaning of a word is by going here and there coincidence or forming again of repetition
Very_well is pretty good; The doubts_of_questions query;
Bear_fruit result; The_far_distance at a distance;
Through the operation of above-mentioned group of compound word, often in a sentence cabin, part speech string has been formed multiple string, makes part complex sentence cabin become the simple sentence cabin.But also have part sentence cabin still complicated, and contain the composition of similar sentence pattern.This will extract the cabin modulo operation.
The operation of above-mentioned " extracting the cabin mould " is complex sentence cabin abbreviation the simple sentence cabin exactly.It comes out similar sentence pattern extracting section as the cabin mould in the face of whole complex sentence cabin (in the following routine brace), and similar sentence cabin part is cabin eye (in the square bracket in the following routine brace).The content of a cabin eye also is to remove some negligible speech strings to be not more than three sense-group text strings outward as the simple sentence cabin.Illustration is as follows again:
Where?have?you?been?since?1{1[I]last?saw?2[you]}？
From 1{ 1[last time I] with 2[you] meet after, you where?
Were?1{we}2{eloquent?as?angels}，yet?1{we}should?please?some?people?3{more?by1[listening]than?2[talking]}.
Even 1{ is we } have 2{ to resemble angelical eloquence }, also still the many 1[of 3{ listen] few 2[says] people is liked.
When1{Marco?Polo}brought?2{them}this?information，3{they}found?it?4{too1[strange]for?2[belief]}.
When vigorous sieve of 1{ Marko } these things tell 2{ they the time, 3{ they feel that 4{1[is strange] must be difficult to 2[and put letter].
4. volume of compartment detects
Volume of compartment detects by more special expert and utilizes " volume of compartment detection system " to carry out.This module is the relevant operation that the content at sentence cabin and cabin eye detects.As shown in Figure 4, by 401 beginnings, the last operation of coming off duty and interrupting that judged whether to continue continuous last time 402 is redirect sign search 411; Be not to change volume of compartment to extract 403,403 to corpus B404 and read (sentence to) by record, and will wherein contained sentence cabin, the content of cabin eye takes out and deposits table of tank capacities 405 in, first, second language is placed on A language volume of compartment, B respectively and speaks on the volume of compartment field; Corresponding each cabin, or the content of each cabin eye is as a record of table of tank capacities 405.So operation comprises also that wherein they all record on the table of tank capacities 405 from the relevant information of corpus B404 physical record till the full content of corpus B404 or whole taking-up of wherein appointed one section content quilt.Carry out a series of detecting operation by 406～409 then, if incongruent situation is made sign 410 and make relevant error flag on table of tank capacities 405 attribute fields.After these a series of detecting operations finish, sign search 411 is by the attribute field on the record searching table of tank capacities 405, whenever searching a record that sign is arranged, submitting to error-correction operation 412 allows the expert carry out error-correction operation targetedly: for example ought make mistakes from corpus B, can revise 414 by the storehouse example and correct corpus B; When makeing mistakes, can correct simultaneously by 415 pairs of sentence pattern storehouses 420 of sentence pattern sentence example and corpus B404 from sentence pattern and sentence example; When makeing mistakes from cabin eye bank (sense-group string storehouse), can be by the speech string in the middle of the school speech string 418 check and correction sense-group string storehouses 422, or fall checking storehouse 417 and look for makeing mistakes in the middle of the sense-group string storehouse 422 a little to correct; Can make amendment by changing mould 416 again, and deposit the cabin mould in mould storehouse, cabin 421 when becoming certain multiple falsification the pattern of cabin mould to be more suitable for when the expert thinks, can also carry out error correction the cabin mould.After the expert had finished sign search 411 and submits to the current operation of specific aim error-correction operation 412, next mark 413 found and judges when next bar has the record of error flag, and 411 submit to 412 with the new record that needs error correction, carry out the aforesaid operations of next round.419 survey when declaring the expert and will come off duty when coming off duty not, redirect finishes 423 and mops up, and this program run finishes.
5. the type example compares index
Above address, when type example just can be considered expert's comparison work is come to an end when reaching 1%; Adding new literary composition plants.Present embodiment is as above-mentioned, after finishing English, Chinese.Can consider to add the 3rd literary composition plants.When after adding new literary composition kind, reaching 1% again equally; Can consider to add the 4th literary composition again plants.Or the like analogize.
6. whole storehouse coding
Work out the operation (above address a model, it is the preparation of sentence pattern sign indicating number, is stored in the sentence pattern code field, leaves out behind the whole storehouse, and unified organizational system is the sentence pattern sign indicating number again) of meaning unifier sign indicating number by the operation of the technical director in the middle of the expert " whole storehouse coded system ".Meaning unifier sign indicating number is nybble fixed length 16 carry system codes.Wherein a left side the 1st～2 byte is high-order, the coding of representative group or class; It is low level that the 3rd～4 byte is played on a left side, in order to each member in group or the class is encoded.Two sign indicating numbers of 0000H, FFFFH high-order, that low level is central are empty, are not equal to the efficient coding that is encoded to of 0000H and FFFFH.As indicated above, depositing word between the different language literal, speech, sentence and expressing the meaning and be able to equity and data and information such as unified sentence pattern and sense-group text strings; What need establishment meaning unifier sign indicating number is four storehouses such as sentence pattern storehouse, sense-group string storehouse, mould storehouse, cabin and little idiom storehouse.Be the sentence pattern sign indicating number wherein towards the sentence pattern storehouse.Every sentence pattern is compiled one yard, and promptly the every record in sentence pattern storehouse is compiled one yard.Represent this sentence pattern and this sentence pattern corresponding sentence pattern framed structure in each literary composition is planted.Wherein towards sense-group string storehouse be the sense-group sign indicating number, each sense-group is compiled one yard, promptly the record in sense-group string storehouse is compiled one yard.Represent a sense-group and this sense-group sense-group text strings in each literary composition is planted.Towards mould storehouse, cabin be cabin mould sign indicating number, each cabin mould is compiled one yard, promptly the record in cabin mould storehouse is compiled one yard.Represent a cabin mould and the corresponding expression of this cabin mould in each languages.Towards the idiom storehouse be the idiom sign indicating number, every little idiom is compiled one yard, the record in promptly little idiom storehouse is compiled one yard.Represent the corresponding expression of this little idiom in each languages.
Meaning unifier sign indicating number is nybble fixed length 16 system codings.Wherein to be assigned with a sign indicating number section high-word be FF00H～FFCFH to the idiom sign indicating number; It is F000H～FEFFH that the sentence pattern sign indicating number is assigned with sign indicating number section high-word; It is EF00H～EFFFH that cabin mould sign indicating number is assigned with sign indicating number section high-word; It is 0001H～EEFFH that the sense-group sign indicating number is assigned with sign indicating number section high-word.In addition, also having new string mark high-word is FFF0H, and its expression user coinage converges puzzled tentative discovery, invention new ideas; Specific term or the like; The meaning of a word generic that first byte representation of its low level should be gone here and there; Second byte is gone here and there sequence number in article for this; Should new string be interior which the new string (the maximum numbering is 256) of article promptly; New string and note thereof are deposited in the file header.Section transformation character (FFF1H 0000H), expression are that the new paragragh of article begins.Sentence cabin sequence number (FFE0H～FFEFH) expression back Sui content is a cabin content, this cabin sequence number that rise on a left side on sentence pattern, sequence number 0～FH, also represent on the sentence pattern sentence cabin number should≤16.Cabin eye sequence number (expression of FFD0H～FFDFH) back Sui content is that the cabin intraocular holds, this cabin eye sequence number that rise on a left side on the mould of cabin, sequence number 0～FH, also represent on the mould of cabin cabin eye number should≤16.
So meaning unifier sign indicating number comprises: sentence pattern sign indicating number, sense-group sign indicating number, little idiom sign indicating number, cabin mould sign indicating number, sentence cabin sequence number, cabin eye sequence number, new string mark and section transformation character etc.
" whole storehouse coded system " begins by 501 as accompanying drawing 5 its flow processs.502 at first to idiom storehouse 503 delete heavily, ordering or the like arrangement property operation, and examine and encode after errorless, their high bit code is FFE0～FFEFH.504 at first to sentence pattern storehouse 505 delete heavily, ordering or the like arrangement property operation, and examine and encode after errorless, their high bit code is F000H～FEFFH.506 at first to mould storehouse, cabin 507 delete heavily, ordering or the like arrangement property operation, and examine and encode after errorless, their high bit code is EF0FH～EFFFH.508 at first to sense-group string storehouse 509 delete heavily, ordering or the like arrangement property operation, and examine and encode after errorless, their high bit code is 0001H～EEFFH.After aforesaid operations coding is finished, by 510 end of winding up.
Below 7～10 joints be the step that practices of present embodiment:
7. utilize meaning unifier sign indicating number to generate the logical text of meaning
Utilizing meaning unifier sign indicating number to generate the logical text of meaning realizes by means of " certain language generates the logical TEXT system of meaning ".Native system belongs to applying portion of the present invention; According to dividing version in order to the literary composition kind that generates the logical text of meaning; The people that offer each literary composition kind generate mother tongue the usefulness of the logical text of meaning.The user is exactly operator, original text author.This flow process is shown in Fig. 6～7, by 601 beginnings of Fig. 6.Read in sentence, it is unit with the sentence that language statement in source reads in 602, reads in sentence of source language at every turn, and this reads in and can read in the traditional text file, also can accept the real-time sentence input of user.Look into idiom then, looking into the idiom process has the support in idiom storehouse 605, search judge current read in the sentence whether be little idiom, if idiom not 603 judgements is little idioms, promptly in the idiom storehouse, find current little idiom, will take out with the little idiom code field content of record then, (what provide with italic prompting is 16 systems, the 4 bytes meaning unifier sign indicating number form of above addressing to provide idiom sign indicating number 604, down together), promptly through 606, contact C is given to the logical text text 722 of generation meaning of Fig. 7.If not little idiom, here be the version that English generates the logical text of meaning example to offer English, beauties, what read in is not little idiom as current, be " the fisherman consents to return the feather suit, on condition thatfairy dance and play heavenly music for him. ".Then change down and join the operation that 607 pairs of source statement of sentence pattern are joined sentence pattern, join the support that sentence pattern has sentence pattern storehouse 608.Find out the sentence pattern that mates with current sentence to the sentence pattern storehouse, institute joins sentence pattern and judges that unique denys 609, if not unique, intervention is selected 610 and accepted user intervention, selects the sentence pattern that mates most; If sentence pattern is unique middle choosing, as with current sentence coupling " the[1] consents to return[2]; on condition that[3]. " pivoted housing sentence pattern down, source language statement is inserted in sentence pattern 611, soon source statement selects sentence pattern in being inserted in, current sentence become " the1[fisherman] consents toreturn 2[the feather suit], on condition that 3[fairy dance and play heavenly music for him]. ".Be inserted in the process, if contrary opinion 612 (when for example occurring two sentence cabins continuously machine can't judge cut-off and produce contrary opinion) is arranged, if exist contrary opinion to determine 613 to accept user intervention, determine by intervening.If there is not contrary opinion, following transit point B provides sentence pattern sign indicating number 701 to Fig. 7's, provides the sentence pattern sign indicating number in generating the logical text text 722 of meaning through 704.Change down then and get a cabin 702, from above-mentioned type example combining form of being inserted in source statement,, take out a sentence cabin successively, as first cabin " 1[fisherman] " by left-to-right; Provide this cabin sequence number 703, promptly be given to through 704 and generate the logical text text 722 of meaning with the double byte code form.Change judging whether simple cabin 705 then down, if current sentence cabin belongs to the simple sentence cabin, following commentaries on classics 712; If not then being the complex sentence cabin, as " 3[fairy dance and play heavenly music for him] "; Join 706 operations of cabin mould, by the support in mould storehouse, cabin 707.After finding the cabin mould, whether unique 708, if not unique, accept user intervention 709 if judging that the cabin mould is joined by institute, select unique accurate cabin mould; If the cabin mould is unique, as " [1] [2] and[3] forhim ", with this cabin content be inserted in the cabin mould " 1[fairy] 2[dance] and 3[play heavenly music] for him " be inserted in the cabin mould and provide this cabin mould sign indicating number 710, cabin mould sign indicating number is given to generates the logical text text 722 of meaning.Take out a cabin eye successively by left-to-right then, and give deliver from vault eye sequence number 711.Change down again to current sentence cabin or when the content of front deck eye one by one vocabulary (sense-group text strings) carry out the meaning of a word and determine 712, the meaning of a word defines sense-group string storehouse 713 and supports that the user can intervene when the meaning of a word need to be determined; The meaning of a word as " fairy " has " fairy maiden, fay, imaginary ", intervenes choosing " fairy maiden " (this is to determine with the meaning of a word that the Chinese is released under the English pattern) here.The meaning of a word provides this sense-group sign indicating number 714 after determining.If the operation that the meaning of a word is determined can not get the support in sense-group string storehouse, promptly nothing is found in this storehouse.Then it is judged as special string 715, they may be proper noun, the sense-group text strings that sense-group string storehouse also is not indexed to, or user author's new term of inventing, creating out.These special strings do not have the sense-group sign indicating number, are made up of source language character string, and by new string mark marker tape, the absolute altitude position double byte of newly going here and there is FFF1H, and the low level double byte is this string length, and its length is the minimum integral multiple of nybble meaning unifier sign indicating number; These are handled in special string manipulation and provide in 716 modules and finish.In the processing of this special string, also accept input user comment 717; If 715 judge it is not special string, connect down and judge current sentence cabin or finish 718 when front deck eye (whether), if also do not finish when front deck eye or sentence cabin, redirect 712 is carried out the meaning of a word of next vocabulary and is determined to operate; If when the front deck eye has finished then looked into the distance commentaries on classics 711 to begin next cabin eye is operated.If finish in current sentence cabin, then descending, judge that sentence finishes 719, do not finish redirect 702 and carry out the next one and get a cabin operation.If this operation finishes, do next and 720 jump to 602 of Fig. 6 by contact A and carry out reading in and operating of next source statement.If the end of the logical text generation of meaning or halt midway finish operation by finishing 721.
Here supplementary notes generate the logical text text 722 of meaning for receiving multiple outputs such as above-mentioned idiom sign indicating number, sentence pattern sign indicating number, sentence cabin sequence number, cabin mould sign indicating number, cabin eye sequence number, sense-group sign indicating number, special string, they are noted, and are (this conform to each languages custom) by the order record that successively takes place.But when current sentence finishes, do next sentence 720 running before, with the interpretation of records of current sentence for number arranging from small to large by sentence cabin number, cabin eye; The meaning of Sheng Chenging is led to text like this, and sentence cabin, cabin eye are that unified ascending order is arranged.Above-mentioned example (just read and give coding title and low word bit decimal number)=" sentence pattern sign indicating number 001059; No. 1, sentence cabin preface; Sense-group sign indicating number 016844; No. 2, sentence cabin preface; Sense-group sign indicating number 023378; No. 3, sentence cabin preface; Cabin mould sign indicating number 00205; No. 1, cabin eye preface; Sense-group sign indicating number 016845, No. 2, cabin eye preface; Sense-group sign indicating number 001802, No. 3, cabin eye preface; Sense-group sign indicating number 017657; Sense-group sign indicating number 016853; Sense-group sign indicating number 005772 ".If " dance for him and play under the condition of melody in the sky the fairy maiden, the fisherman promises to give back the garment or robe made of feathers with the Chinese sentence of identical semanteme again." generate to anticipate and lead to text sentence=" sentence pattern sign indicating number 001059; No. 1, sentence cabin preface; Sense-group sign indicating number 016844; No. 2, sentence cabin preface; Sense-group sign indicating number 023378; No. 3, sentence cabin preface; Cabin mould sign indicating number 00205; No. 1, cabin eye preface; Sense-group sign indicating number 016845, No. 2, cabin eye preface; Sense-group sign indicating number 001802, No. 3, cabin eye preface; Sense-group sign indicating number 017657; Sense-group sign indicating number 016853; Sense-group sign indicating number 005772 "; As can be seen, it is just the same with the code that English generates the logical text of meaning.The logical text of meaning that Here it is, the feature of same peculiar disk file.
8. the logical text of meaning reads
Reading by means of " the logical text language of meaning read-out system " of the logical text of meaning realizes.This system also belongs to applying portion of the present invention, divides version with the literary composition kind of reading, and the people that offer each literary composition kind read the usefulness that meaning is led to text, and the user is exactly the operator.The logical text generation of the logical text read-around ratio meaning of meaning is simple, thereby earlier version just can offer domestic consumer's use.Native system flow process such as accompanying drawing 8.By 801 beginnings, read sentence operation 802, it reads a logical text sentence of meaning to the logical text 803 of meaning.This sentence can be a little idiom (only being a little idiom sign indicating number); It also can be a long sentence (the current sentence of expression finishes before next sentence pattern sign indicating number or little idiom sign indicating number) that carries various meaning unifier sign indicating numbers by the sentence pattern sign indicating number.If 802 what read in is an idiom sign indicating number, get little idiom immediately, handle 804 by little idiom, from little idiom storehouse 805, take out this little idiom, and should little idiom send and show and export 806, send file standby 807 simultaneously.If 802 what read in is not a little idiom but sentence that is carried by the sentence pattern sign indicating number is then inserted generation buffer memory 808 with whole sentence, for example inserted " sentence pattern sign indicating number 001059; No. 1, sentence cabin preface; Sense-group sign indicating number 016844; No. 2, sentence cabin preface; Sense-group sign indicating number 023378; No. 3, sentence cabin preface; Cabin mould sign indicating number 00205; No. 1, cabin eye preface; Sense-group sign indicating number 016845, No. 2, cabin eye preface; Sense-group sign indicating number 001802, No. 3, cabin eye preface; Sense-group sign indicating number 017657; Sense-group sign indicating number 016853; Sense-group sign indicating number 005772 ".Get sentence pattern then, by the following part step-by-step processing: sentence pattern operation 809 is read sentence pattern sign indicating number 001059 in the middle of generation buffer memory 808; Read this statement type (reading version with Chinese is that example is the content of " Chinese type " field) by sentence pattern storehouse 810, provide 811 by sentence pattern, be given to a buffer memory 812, this sentence pattern is " under the condition of [3], [1] promises to give back [2] ".The professional is not difficult to understand in the row, yet following operation is exactly to decipher one by one, takes out corresponding contents to the Chinese respective field in corresponding storehouse, and the corresponding sentence cabin that is inserted into this sentence pattern gets final product.Its flow process is that a cabin is from left to right searched in definite current sentence cabin 813 successively, determines current sentence cabin; Get a cabin then, read all codes in this number sentence cabin in the generation buffer memory 808; From left to right handle by sign indicating number.Then, if 814 determine thereafter be cabin mould sign indicating number; Get cabin mould 815 and utilize this cabin mould sign indicating number in mould storehouse, cabin 816, to read this cabin mould, and this cabin mould is placed a cabin buffer memory 817, then in the middle of front deck eye 818 determines to exist sentence cabin buffer memorys when front deck eye (being from left to right to search for equally); If 814 what declare is not cabin mould sign indicating number, jump to and judge sense-group string 819, as to determine be sense-group string code, then taking-up is inserted 820 and is arrived sense-group string storehouse 821 these sense-group text strings of taking-up, through 822 indication current sentence cabin or cabin eyes, be the current position that should insert, this sense-group text strings is inserted into appropriate location among a buffer memory 812 or the sentence cabin buffer memory 817.Betraing to such an extent that be not meaning unifier sign indicating number if judge sense-group string 819, give and judge special string 823, is after suitably being handled by special string manipulation 824, is inserted into the position that insert through 822 indications.By judging, finish to jump 819 and continue then when front deck eye end 825; Finished again by judging whether that current sentence cabin finishes 826, finished to jump to revise when the front deck eye and proceed when front deck eye 818; If finish, arrive sentence again and finish 827, sentence does not finish to jump 813 and carries out the cabin operation of next one sentence; If sentence finishes to judge whether to need to suspend 828 again, do not need time-out to jump 802 and read in next sentence, carry out the operation of next sentence.In the middle of the insertion of above-mentioned distich cabin or cabin eye, also have caching management module 829, its can be incorporated them among 812 current cabins of a buffer memory when sentence cabin buffer memory 817 is expired, and empties 817; Expired and be sentence when finishing when 812, it can be sent to 812 content 806 and 807 and empty 812 arrivals in order to next sentence again.Simultaneously 829 also have word order to intervene 830 supports, and when simple sentence cabin or cabin eye provide when finishing, they might need levelling by word order, and word order intervention at this moment 830 automatically or accept the word order levelling that user intervention carries out current sentence cabin or cabin eye.829 supports that also have loss compensation 831 in addition, can plant certain literary composition may not need manual intervention maybe to need the compensation of intervening by lost part when generating the logical text of meaning, for example Ying Wen article, Chinese measure word, and refer to, the abridged fills up or the like.Above-mentionedly judge whether to suspend at 828 o'clock, suspend if desired and then make tailing in work 832, native system end of run.
9. stride the languages AC network
Stride languages AC network system as shown in Figure 1.101 presentation surfaces may operate on the personal computer of unit or networking to a languages version software of Chinese language users among the figure.Wherein contain these languages read apparatus and one-writing system (for example Chinese generates logical TEXT system of meaning and Yi Tong text Chinese read-out system), other languages by that analogy.102～108 also as 101 among the figure, and presentation surface is to the languages version of English, Russian, Spanish or the like languages respectively.109～112 difference presentation surfaces are to the corresponding languages reading and writing of other languages user system among the figure.113 expressions connect into net with 101～112 among the figure; Wherein also connect and have a mind to lead to text server 114～115; And stride languages and exchange service centre 116; Their common compositions are striden languages AC network system.Each languages version is towards the customer group of corresponding languages, and this group user utilizes mother tongue to read and write the logical text of meaning.The logical text of the meaning of being write as can be stored on the memory of user personal computer, also can be stored on the webserver, also can transmit between whole network, communication.No matter, when other user search needs to it, just become his mother tongue when reading by his personal computer from the logical text of the meaning of what languages.Yet the people of different language can utilize mother tongue separately to read and write, exchange.
Striding on the languages AC network, be provided with and stride languages and exchange service centre, this center at least one.Stride languages and exchange service centre's groundwork: A, collect string especially, they are collected, put into different categories, improve its concept definition and write down frequency.Include new string storehouse in, accept user's consulting and inquiry.After the test through certain hour and the amount that is used, perhaps be eliminated; Perhaps be included into sense-group string storehouse.Languages AC network system is striden in B, upgrading, and in sum, the logical text of meaning comes from the comparison arrangement of expert to the sample language material.But the upgrading of the logical TEXT system of meaning does not need the sample language material, directly utilizes the logical text statistics of meaning, analyzes; Because the logical text of meaning itself is good sample language material.C, user support, gives many-sided support according to the user is required.D, guide multilingual common development, stride the appearance that languages exchange service centre, the history that the natural language that just is through with develops lentamente and develops in independent system separately; Begun the course of multilingual common fast development.For example to revise, eliminate or newly-increased sense-group text strings; Promoting new terminology or the like can be by striding the languages AC network directly to writer's suggestion, recommendation; Propagate explanation to the reader.
10. another embodiment
In addition to the implementation, according to design of the present invention, more embodiment can also be arranged.For example:
The foreign language system read in mother tongue
The numerous people that need the foreign language article of tradition of reading towards being ignorant of foreign language provide " the foreign language system read in mother tongue ".Native system belongs to applying portion of the present invention, offers the usefulness that a blind user of the also unacquainted foreign language of foreign language word utilizes the mother tongue reading foreign language.Read in the process of foreign language at mother tongue, " the foreign language system read in mother tongue " not only provides reader's mother tongue; Also generate the logical text of meaning simultaneously.Yet as long as one piece of traditional article has a people to read, thousands upon thousands people of back just can become the logical text of reading meaning.Communicating with the eyes of native system is user's mother tongue, and the source language is certain foreign language beyond this mother tongue.Here might as well be mother tongue with Chinese, English be that the source language is described as follows for example, and flow process is shown in Fig. 9～10.By 901 beginnings of Fig. 9, outer sentence, promptly source statement reads in 902 and reads in source language (foreign language) sentence.Declare idiom then and not 903 do not declare its whether little idiom (promptly look into and contain in the idiom storehouse not), provide the little idiom of order idiom 904 inquiries storehouse 905, find, take out the corresponding little idiom (this example is " middle habit " field contents) of target language, female little idiom directly is given to demonstration output 906, also is given to and files standby 907; In this 904～907 process, also provide the exhausted logical text of meaning that generates simultaneously of this idiom sign indicating number and use.If 903 judge it is not little idiom, for example are " the tragedy is not in notknowing, but in not knowing that you do not know. "; Join sentence pattern 908, join optimum matching promptly searched in sentence pattern storehouse 909 (this example is an English type-word section) in sentence pattern with source statement sentence pattern, take out the sentence pattern of corresponding communicating with the eyes (medium-sized field) simultaneously.They are " [1] is not[2], but[3]. "; " [1] is not [2], but [3]."; Judge that then whether unique this sentence pattern 910,, intervene and select 911 interventions of accepting the user, select best coupling if not unique; If 910 judge that this sentence pattern is unique, parent form provides, mother tongue (this example is medium-sized) sentence pattern is directly provided: under continuous source statement is inserted in source statement type 912, as " 1[thetragedy] is not 2[in not knowing], but 3[in not knowing that you do not know]. "; If determine and be inserted in sentence pattern contrary opinion 913 (a sentence cabin for example occurs continuously, computing machine is incompetent determines cut-off) is arranged, can intervene and determine that 914 accept user intervention and determine cut-off.If 913 judge there is not contrary opinion, provide the sentence pattern 915 of communicating with the eyes (also provide simultaneously this sentence pattern sign indicating number be equipped with generate the logical text of meaning with) in the buffer memory 916 of communicating with the eyes, as " [1] is not [2], but [3]."; Get a cabin 917 then, from left to right search for the central sentence cabin of the sentence pattern of communicating with the eyes, and in source statement, take out corresponding sentence cabin content with the label in this cabin.Judge again whether this cabin content is simple cabin 918, if C is descending for the redirect contact; Here 1～2 two sentence cabin all is the simple sentence cabin; 1[the tragedy] 2[in_not_knowing]; If not, be the complex sentence cabin as the 3rd sentence cabin, 3[in_not_knowingthat you do not know].Join cabin mould 919 then, select the cabin modulo operation according to the support in mould storehouse, cabin 920.If it is unique 921 to determine selected cabin mould, accept user's intervention 922; If it is unique that institute is declared, " [1]+that you do not know "; " you do not know+[1] ".Cover cabin mould 923 is inserted in the cabin mould to source this cabin content of speaking as " 1[in_not_knowing] that you donot know "; Simultaneously the cabin mould of communicating with the eyes is deposited among the cabin buffer memory 924.So far Fig. 9 indicates A, B, C, D, five contacts of E, and is continuous mutually by the corresponding A of Figure 10, B, C, five contacts of D, E respectively.After 923 finish cover cabin mould, go downwards to Figure 10 meaning of a word of communicating with the eyes by contact C and determine 1001, it has the support in sense-group string storehouse 1002, also can accept user's intervention in case of necessity; Can be chosen by the user if run into the ambiguity string, provide the text strings 1003 of communicating with the eyes, it is by current sentence cabin or when the indication 1004 of front deck eye, again by contact D, this sense-group text strings (the speech string of communicating with the eyes) is inserted among the sentence cabin buffer memory 924 or sentence buffer memory 916 of Fig. 9.This example 1[the tragedy] meaning of a word is " tragedy in this world "; " in_not_knowing " is multiple string, and the meaning of a word is " ignorant "; Fill out a cabin, they inserted corresponding sentence cabin or cabin eye respectively, this example become " 1{ is tragedy in this world } be not 2{ ignorant }, but 3{ you do not know 1[ignorant]."; On the other hand, 924,916 also accept the control of cache management 1009, finish as current sentence cabin, include sentence cabin cache contents in buffer memory, empty a cabin buffer memory; When full sentence finishes, its can be given to the content of sentence buffer memory and show output 906 and file standbyly 907, and empties the arrival that a buffer memory 906 is met new sentence.In aforesaid operations, provide the text strings 1003 of communicating with the eyes and also provided this meaning unifier sign indicating number; And also they are stored in the lump when the user selects the meaning of a word, gravity treatment such as ambiguity string are equipped with and regret 1004 when regretting.If 1001 can not determine the meaning of a word of communicating with the eyes, unregistered word bar etc. just appears, judge that it is special string 1006, provides special string 1007.Judge again whether current sentence cabin or cabin eye finish 1008,, jump 1001 and continue the operation that next speech string is determined the meaning of a word if 1008 determine current sentence cabin or the cabin eye does not finish; If finish, judging whether that sentence finishes 1010.Do not take out next sentence cabin if finish to jump 917 of Fig. 9, carry out the operation in next sentence cabin through contact B.If sentence finishes, judging whether needs to suspend 1011 again, if do not suspend, jumps 902 of Fig. 9 through contact A and reads source language next sentence, a wheel operation again.In addition, above-mentioned cache management 1009, it also has word order to intervene 1013 support, when the entry of communicating with the eyes of simple sentence cabin or cabin eye all provides, may need the word order intervention, at this moment 1013 can accept user intervention.Cache management 1009 is regreted the support of operation, certain language loss compensation 1014 in addition, can mend the possible drop-out of certain literary composition kind and taste; When regretting operating together with the ambiguity string be equipped with to regret 1004, the buffer memory 916 of communicating with the eyes, sentence cabin buffer memory 924 and cache management 1009 preparation in advance for information about, user's operation that the meaning of a word determines that choosing is staggered the time and can be repented and mended one's ways of regretting communicating with the eyes.Above-mentionedly judge whether to suspend 1011, suspend if desired, then make tailing in work by stopping 1012, native system runs abort.
Above-mentioned example is whole flow process process; The variation step by step of source language can provide when programming was implemented, and also can conceal, and made the user optional; Next is that full sentence provides or minute sentence pattern, provides and also allow the user optional to a sentence cabin, a sentence cabin.Can adapt to slowly needs reference or the like the requirements of different users of reading, read soon or learning a foreign language like this.
11. isolated plant
Implement the isolated plant of the inventive method, as shown in Figure 11.It can be made into as intercom, mobile phone, digital telephone, personal digital assistant, and the device of earphone or the like form.The scientific seminar that is used for not exchanging between the identical text kind, exchanging meeting, tourism or the like directly utilize the occasion of oral communication.This device is planted with the fixed literary composition of model to divide model towards the people of identical text kind not, and each user selectes type with own mother tongue.
The functional structure of this device is divided two flow processs:
1. the user says flow process
When the user said, user's voice entered parts phonetic entry 1102 by microphone, did not have voice to enter in advance, signaling switch of this vice activation that has just entered, also shutdown signal reception 1101 when opening phonetic entry.The user speak during (during also promptly having voice to send continuously), phonetic entry 1102 work becomes the mother tongue character code to speech conversion.The mother tongue character code enters mother tongue sentence parts 1105 one by one, and its forms sentence to character string of full sentence, and according to processings of correcting mistakes of the sentence pattern (if wrong) of mother tongue, then correct mother tongue sentence is passed to mother tongue and is generated to anticipate and lead to text parts 1106.These parts lead to code string to the meaning that the mother tongue sentence converts the logical text sentence form of meaning to, and to signal emission element output code string.When signal emission element 1103 receives code string, the logical code string of meaning is converted to digital signal corresponding, and launch.This has just finished the flow process that the user says.
2. the user listens flow process
When after finishing in short, stopping, be in a platform that does not have voice signal when longer, this platform starts a signaling switch again, has opened signal and has received 1101 and close phonetic entry 1102 simultaneously.Signal receives 1101 beginning receiving digital signals, and digital signal is converted to meaning unifier sign indicating number, and the unifier of will anticipating sign indicating number is passed to the logical text mother tongue of meaning and read 1104 parts.These parts unifier sign indicating number of will anticipating converts mother tongue sentence pattern, mother tongue character string such as sentence cabin etc. one by one to, and the latter is passed to mother tongue sentence parts 1105.These parts are integrated into the mother tongue sentence with these mother tongue character strings, then phonetic synthesis output block 1108 are passed in the mother tongue sentence.These parts carry out the phonetic synthesis of this mother tongue sentence, voice output then, and the user just can hear the other side's language (hearer's mother tongue).The flow process that this user listens finishes.
The logical text mother tongue of meaning is read, mother tongue sentence, mother tongue generate logical text three parts 1104～1106 of meaning and mainly be made up of single-chip microcomputer, memory and be connected with USB interface, prepares against and changes data from computing machine or network.
The logical text mother tongue read-out element 1104 of meaning is the parts that can move " the logical text language of meaning read-out system " step in the inventive method.
Mother tongue generates the logical text parts 1106 of meaning, is the parts that can move " certain language generates the logical TEXT system of meaning " step in the inventive method.
Here " meaning logical text language read-out system " of operation or " certain language generates the logical TEXT system of meaning " comprised contains intentionally the storehouse of unifier sign indicating number and all gets succinct branch storehouse form.And can further divide the scope of application according to daily life, tourism, specialty etc., together with above divide model with the literary composition kind, form the type class of difference in functionality price ratio jointly.
1101, emission between 1103 parts, the signal that receives are the digital signals of the satisfied unifier sign indicating number of the inventive method.Also can receive 1101 to signal, signal launches and changes into wiredly between 1,103 two parts, is used for the wire communication occasion and the device product thereof of numeral.
Software systems of the present invention can be moved enforcement on device provided by the invention; Also can be at existing medium-sized, small-sized, little, supercomputer, notebook computer, palm PC etc. independent or the computing machine into the net that links to each other on operation implement; Can particularly implement at various computer networks in operation on the Internet; Can also move on the device of PDA (Personal Digital Assistant) and implement such as " personal digital assistant ".Soft, hardware product after the invention process, the work that can be applied to need to exchange, study, leisure, tourism or the like occasion with the people of other languages; Can be used for the occasion that family, office, school and all trades and professions relate to foreign language.
Claims (10)
1, a kind of text-converted method is characterized in that, comprises the steps:
S1, the reception first natural language text file, the meaning of calling the meaning of one's words of the corresponding described first natural language text file of first natural language storehouse generation is led to text;
S2, reception second natural language output command are called the second natural language storehouse, generate according to the logical text of described meaning and also export the second natural language text file.
2, text-converted method according to claim 1 is characterized in that, described first language storehouse comprises the first sentence pattern storehouse and storehouse, first cabin, and described step S1 further comprises the steps:
S11, read in a sentence in the described first natural language text file;
S12, call the described first sentence pattern storehouse sentence that reads in is carried out sentence pattern coupling, output sentence pattern meaning of one's words code, and mark off the sentence cabin of the described sentence that reads in;
S13, call storehouse, described first cabin and cabin coupling is carried out in a sentence cabin that marks off among the step S12, an output sentence cabin meaning of one's words code;
S14, the described sentence pattern meaning of one's words code of foundation and described sentence cabin meaning of one's words code generate the meaning unifier sign indicating number of the corresponding described sentence that reads in;
S15, judge whether to arrive the described first natural language text end of file, if execution in step S16 if not, then reads in next sentence, gets back to step S12 then;
The meaning unifier sign indicating number of S16, all sentences that read in that will generate is stored as the logical text of described meaning.
3, text-converted method according to claim 2, it is characterized in that, the described first natural language storehouse comprises the first idiom storehouse, in described step S12,, call the described first idiom storehouse and judge whether the described sentence that reads in is idiom calling the described first sentence pattern storehouse carries out sentence pattern coupling to the sentence that reads in before, if, then export idiom meaning of one's words code, the meaning unifier sign indicating number as the described sentence that reads in forwards step S15 then to; If not, then call the described first sentence pattern storehouse sentence that reads in is carried out the sentence pattern coupling.
According to claim 2 or 3 described text-converted methods, it is characterized in that 4, storehouse, described first cabin comprises the mould storehouse, first cabin and the first cabin eye bank, described step S13 further comprises:
S131, read in a sentence cabin of the described sentence that reads in;
S132, call mould storehouse, described first cabin cabin mould coupling is carried out in a sentence cabin of reading in, if the match is successful, then export cabin mould meaning of one's words code, and mark off the cabin eye in described sentence cabin of reading in, forward step S133 then to; If it fails to match, then forward step S134 to;
S133, call the described first cabin eye bank cabin eye that marks off is carried out cabin eye coupling, output cabin eye meaning of one's words code, and generate the sentence cabin meaning of one's words code in corresponding described sentence cabin of reading in according to described cabin mould meaning of one's words code and described cabin eye meaning of one's words code, forward step S135 to;
S134, call the described first cabin eye bank cabin eye coupling is carried out in a sentence cabin of reading in, output cabin eye meaning of one's words code, the sentence cabin meaning of one's words code as described sentence cabin of reading in forwards step S135 to;
S135, judge whether to exist untreated sentence cabin, if, then read in next sentence cabin, get back to step S131 then, otherwise, forward step S14 to.
5, text-converted method according to claim 4, it is characterized in that, in described step S133 or S134, if eye coupling in described cabin is unsuccessful, then in the described first cabin eye bank, look unfamiliar into cabin eye meaning of one's words code, export this cabin eye meaning of one's words code then for the cabin in the unsuccessful cabin of coupling.
6, text-converted method according to claim 5 is characterized in that, described step S14 further comprises:
Signal selected in S141, reception first sentence pattern, select the sentence pattern meaning of one's words code of corresponding described first sentence pattern selection signal from the sentence pattern meaning of one's words code of output;
Signal is selected in S142, first cabin of reception, selects described first cabin of correspondence to select the sentence cabin meaning of one's words code of signal from the sentence cabin meaning of one's words code of output;
Sentence pattern meaning of one's words code that S143, foundation are selected and sentence cabin meaning of one's words code generate the meaning unifier sign indicating number of the corresponding described sentence that reads in.
7, text-converted method according to claim 1 is characterized in that, described second language storehouse comprises the second sentence pattern storehouse and storehouse, second cabin, and described step S2 further comprises the steps:
S21, read in the meaning unifier sign indicating number of a corresponding sentence in the logical text of described meaning;
S22, call the described second sentence pattern storehouse, convert the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number to corresponding sentence pattern;
S23, call storehouse, described second cabin, convert the sentence cabin meaning of one's words code in the described meaning unifier sign indicating number to corresponding sentence cabin;
S24, according to sentence pattern that generates and sentence cabin generate and the corresponding described meaning unifier sign indicating number that reads in of output sentence;
S25, judge whether to arrive the end of the logical text of described meaning, if not, then read in the meaning unifier sign indicating number of corresponding next sentence, get back to step S22 then.
8, text-converted method according to claim 7, it is characterized in that, the described second natural language storehouse comprises the second idiom storehouse, in described step S22, before calling the described second sentence pattern storehouse, described method also comprises calls the described second idiom storehouse, judge whether described meaning unifier sign indicating number is idiom meaning of one's words code, if, then export the idiom of corresponding described idiom meaning of one's words code, forward step S25 then to; If not, then call the described second sentence pattern storehouse,, generate corresponding sentence pattern according to the sentence pattern meaning of one's words code in the described meaning unifier sign indicating number.
9, according to claim 7 or 8 described text-converted methods, it is characterized in that, storehouse, described second cabin comprises the mould storehouse, second cabin and the second cabin eye bank, described step S23 comprises that further calling described storehouse, second cabin and the described second cabin eye bank converts cabin mould meaning of one's words code and cabin eye meaning of one's words code in the meaning of one's words code of described sentence cabin to cabin mould and cabin eye, and looks unfamiliar into described sentence cabin according to described cabin mould and cabin.
10, text-converted method according to claim 9 is characterized in that, described step S24 further comprises:
Signal selected in S241, reception second sentence pattern, select the sentence pattern of corresponding described second sentence pattern selection signal from the sentence pattern that generates;
Signal is selected in S242, second cabin of reception, selects described second cabin of correspondence to select the sentence cabin of signal from the sentence cabin that generates;
Sentence pattern that S243, foundation are selected and sentence cabin generate and export the sentence of corresponding described second language.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101244658A CN101436179A (en) | 2007-11-13 | 2007-11-13 | Method and apparatus for converting text |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101244658A CN101436179A (en) | 2007-11-13 | 2007-11-13 | Method and apparatus for converting text |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101436179A true CN101436179A (en) | 2009-05-20 |
Family
ID=40710620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2007101244658A Pending CN101436179A (en) | 2007-11-13 | 2007-11-13 | Method and apparatus for converting text |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101436179A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102236645A (en) * | 2010-05-06 | 2011-11-09 | 上海五和际软件信息有限公司 | Semantic logic-based pseudo-natural language human-computer dialogue device |
| CN106383819A (en) * | 2016-01-11 | 2017-02-08 | 陈勇 | Speech convertor |
| CN111931508A (en) * | 2020-08-24 | 2020-11-13 | 上海携旅信息技术有限公司 | Digital conversion method and system, text processing method and system, device and medium |
-
2007
- 2007-11-13 CN CNA2007101244658A patent/CN101436179A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102236645A (en) * | 2010-05-06 | 2011-11-09 | 上海五和际软件信息有限公司 | Semantic logic-based pseudo-natural language human-computer dialogue device |
| CN106383819A (en) * | 2016-01-11 | 2017-02-08 | 陈勇 | Speech convertor |
| CN111931508A (en) * | 2020-08-24 | 2020-11-13 | 上海携旅信息技术有限公司 | Digital conversion method and system, text processing method and system, device and medium |
| CN111931508B (en) * | 2020-08-24 | 2023-05-12 | 上海携旅信息技术有限公司 | Digital conversion method and system, text processing method and system, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11645547B2 (en) | Human-machine interactive method and device based on artificial intelligence | |
| CN109086408A (en) | Document creation method, device, electronic equipment and computer-readable medium | |
| CN101183281B (en) | Method for inputting word related to candidate word in input method and system | |
| CN102622342B (en) | Intermediate family of languages system, intermediate language engine, intermediate language translation system and correlation method | |
| Papegaaij | Word expert semantics: an interlingual knowledge-based approach | |
| CN104462057B (en) | For the method and system for the lexicon for producing language analysis | |
| CN102272755A (en) | Method for semantic processing of natural language using graphical interlingua | |
| CN101246474B (en) | Method for reading foreign language by mother tongue based on sentence component | |
| CN101510194B (en) | A kind of multilingual professional translation method based on sentence component | |
| CN110929007A (en) | A power marketing knowledge system platform and application method | |
| CN116432605B (en) | Composition comment generation method and device integrating priori knowledge | |
| GB2604317A (en) | Dialogue management | |
| RU2688758C1 (en) | Method and system for arranging dialogue with user in user-friendly channel | |
| CN118170919B (en) | A method and system for classifying literary works | |
| CN101436179A (en) | Method and apparatus for converting text | |
| CN108615529A (en) | A kind of control method, device and domestic robot based on semantics recognition | |
| CN118246428A (en) | Recall record generating system based on digital human real-time interaction and generating artificial intelligence | |
| CN103218353B (en) | Mother tongue personage learns the artificial intelligence implementation method with other Languages text | |
| Ahmad | People centered HMI’s for deaf and functionally illiterate users | |
| CN114970733A (en) | Corpus generation method, apparatus, system, storage medium and electronic device | |
| CN1553381A (en) | Multi-language correspondent list style language database and synchronous computer inter-transtation and communication | |
| CN112115722A (en) | Human brain-simulated Chinese analysis method and intelligent interaction system | |
| CN1296588A (en) | Open holographic template type man-machine conversation language translation method and holographic semantic labeling system | |
| Amezian et al. | Training an LSTM-based Seq2Seq model on a Moroccan biscript lexicon | |
| An | Construction and Application of Corpora in Computer-Assisted Translation of Spanish Intangible Cultural Heritage for External Promotion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20090520 |