JP2001005814A - Device and method for preparing dictionary, computer readable recording medium recording dictionary preparation program and translation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a dictionary preparing device capable of more flexibly dealing with the request of a user and a translation device using a prepared dictionary. SOLUTION: A document mixing plural kinds of characters is inputted from an input part 1. A character string extracting part 4 extracts a character string composed of a specified kind of character strings from the input document. A dictionary information preparing part 17 prepares dictionary data by adding corresponding dictionary information such as translated word or part of speech with each extracted character string as an index word and registers these data in a data storage part 3. The dictionary data are displayed through an output part 2. When preparing the dictionary information, a dictionary is prepared by inputting information for dictionary preparation desired by the user through a user supporting part 22 as needed. Therefore, this device is provided for dictionary preparation efficient for the user and translation using the prepared dictionary.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a dictionary creating apparatus for creating dictionary data referred to for translating an input original sentence and outputting a translated sentence, and a translation apparatus for performing a translation process using the created dictionary data. In particular, the present invention relates to a dictionary creation device, a dictionary creation method, a translation device, and a computer-readable recording medium that records a dictionary creation program that can flexibly respond to user requests.

[0002]

2. Description of the Related Art In recent years, a large number of translators using electronic computers have been developed, but the translation results have not yet reached the level of human translation. As one of the causes, there is a problem that dictionary data for translation cannot correspond to proper nouns that increase every day. Such proper nouns are
In many cases, it is sufficient to output the translation result as it is in the language character string. Also, with the internationalization, there is a phenomenon that the concept of one language is used as it is in one language, and the other language is treated as if it were a word or idiom in one language. When such a word or idiom appears in the original text to be translated, it is often easier to read if the word or idiom is output as it is as a translation of the translation result than a directly translated word.

[0003] Further, it is not easy to discriminate words and idioms in one language in the other language when the language before translation and the language after translation are composed of the same character type. In the case of English using "alphabetic characters" and Japanese using "kanji, hiragana, katakana", it is easy to identify using the type of character.

[0004] As a translation apparatus using such a character type, there is a machine translation apparatus disclosed in Japanese Patent Application Laid-Open No. Hei 8-171564. This publication discloses a method of realizing a machine translator (particularly, a machine translator that translates Japanese into another language) that can use the alphabet as a translated word when an unknown word of an alphabet character exists in the original sentence. .

[0005]

However, the technique disclosed in Japanese Patent Application Laid-Open No. HEI 8-171564 only discriminates the character type in the input original text for machine translation. No application to other processes is shown.

In some cases, a document that has been created manually has a higher utility value irrespective of machine translation. For example, manually
If the English sentence x yyy zzz is adopted in the Japanese sentence, the English character string “xx yyy zz”
When translating z ″ into Japanese, it indicates the user's intention that the English character string as it is (the original character string) is desirable or sufficient, so it is translated by English-Japanese machine translation. In such a case, an English character string according to the user's intention can be used.Especially, although many documents can be obtained in a digitized form using the Internet or the like, it has been digitized. Documents can be easily subjected to automatic translation processing, and words can be accumulated with a small load, which can produce a great effect.

Further, Japanese Patent Application Laid-Open No. Hei 8-171564 does not mention much about the method of extracting a character string from a document. However, in practice, the high extraction accuracy and the degree of freedom of user customization are the result of translation. Influences. In some cases, the user may arbitrarily change the information on the extraction accuracy to increase the translation accuracy.

[0008] In order to provide a high degree of freedom in extracting character strings, it is also desired to be able to identify character strings in a third language other than alphabetic character strings. For example, an Arabic character string appears in a Japanese document, and it is desired that the same character string be directly expressed in an English document as a translation result.

[0009] Therefore, an object of the present invention is to provide a dictionary creation device, a dictionary creation method, a translation device, and a computer-readable recording medium that records a dictionary creation program, which can flexibly respond to the needs of the user. That is.

[0010]

A dictionary creating apparatus according to an aspect of the present invention is used for translating a first language represented by a first type of character into a second language represented by a second type of character. A dictionary creation device that creates dictionary information in which unregistered headwords are registered in an existing dictionary used for translation, and for inputting information including a document in which a plurality of types of characters are mixed. A character string extracting unit that extracts one or more character strings of a first type of character from a document input from the input unit and outputs extraction information in which the extracted character strings are registered; A dictionary information creation unit for creating a dictionary information by using the extracted character string of the extracted information output from the column extraction unit as a headword and adding information on the headword as headword information corresponding to each of the headwords; It is comprised including.

According to the above-described dictionary creation apparatus, a character string composed of the first type of character is extracted from a document in which a plurality of types of characters are mixed, and dictionary information registered as a headword is created. The document of the first language indicated by the character of the second
Dictionary information in which an unregistered headword is registered is created in an existing dictionary that is referred to for translation into a second language document indicated by a type of character.

[0012] Therefore, even in the text of the second type language, a term (new word, technical term) that is intentionally described in the first type language, and an unregistered entry word is registered in the existing dictionary. The dictionary information can be created so as to refer to the dictionary information when translating a first language document represented by a first type of character into a second language document represented by a second type of character. If this is the case, more accurate translation processing can be performed.

A dictionary creation device according to another aspect of the present invention is used for translating a first language represented by a first type of character into a second language represented by a second type of character,
An input unit for inputting information including a document in which a plurality of types of characters are mixed, wherein the existing dictionary used for translation creates dictionary information in which unregistered headwords are registered, A character string extracting unit that extracts one or more character strings of a first type of character from a document input from the input unit and outputs extraction information in which the extracted character strings are registered; A dictionary information creating unit for creating dictionary information by using an extracted character string of the extracted information as a headword and providing information on the headword as headword information corresponding to each of the headwords; And an information correction unit for correcting the contents of the dictionary information created by the above as desired in accordance with information input from the outside via the input unit.

According to the above-described dictionary creation apparatus, a character string composed of a first type of character is extracted from a document in which a plurality of types of characters are mixed, and dictionary information registered as a headword is created. The document of the first language indicated by the character of the second
Dictionary information in which an unregistered headword is registered is created in an existing dictionary that is referred to for translation into a second language document indicated by a type of character. Also, the created dictionary information can be changed by the information correction unit from the outside according to information given by the user, for example, as desired by the user, so that user-customized dictionary information or highly accurate dictionary information can be easily obtained. be able to.

[0015] Therefore, even in the text of the second type language, a term (new word, technical term) that is intentionally described in the first type language, and an unregistered entry word is registered in the existing dictionary. The user can customize the dictionary information and create a second language document indicated by the first type of characters.
When translating into a second language document represented by
By referring to the dictionary information, more accurate translation processing can be performed.

In the above-described dictionary creation device, the information correction unit corrects the contents of the dictionary information via the input unit as desired. Therefore, the user can directly modify the contents of the dictionary information via the input unit as desired, and can customize the dictionary information.

In the above-described dictionary creating apparatus, the character string extracting section includes an adjustment processing section for adjusting the extraction of the character string so that the extracted character string is appropriate to be registered as a headword in the dictionary information. It is composed.

According to this dictionary creation device, the headword registered in the dictionary information by the adjustment processing unit is appropriate. Therefore, the contents of the dictionary information have higher accuracy.

In the above-described dictionary creation device, the adjustment processing section is provided with a parenthesis symbol processing section for processing a character string before and after the parenthesis symbol in the extracted character string.

According to this dictionary creation device, even if an extracted character string includes a parenthesis symbol, the character string before and after the parenthesis symbol is processed so as to be adjusted to be a headword.

Therefore, the extracted character string after the included parenthesis symbols have been processed can be registered in the dictionary information as a headword, and the accuracy of the registered headword and thus the accuracy of the dictionary information can be improved.

In the above-described dictionary creation device, the adjustment processing unit is provided with a specific symbol division processing unit that divides an extracted character string before and after a specific symbol in the extracted character string to obtain an individual extracted character string. You.

According to this dictionary creation device, for example, '/'
If a specific symbol that divides a character string such as, is included in the extracted character string, the extracted character string is divided by this specific symbol to be an individual extracted character string.

[0024] Therefore, since a headword containing such a specific symbol is not registered in the dictionary information, the accuracy of the dictionary information is improved.

In the above-described dictionary creation device, the adjustment processing section is provided with a number string removal processing section for removing a number string in an extracted character string.

According to this dictionary creation device, the digit string of the extracted character string is removed. Therefore, for example, a plurality of headwords in which only numbers in a numeral string indicating product version information are different and other character strings are the same are registered in the dictionary information, and the capacity of the dictionary information becomes unnecessarily large. Is avoided. Also, the versatility of the headword is improved.

In the above-described dictionary creating apparatus, the adjustment processing unit includes a layout information processing unit that divides the extracted character string based on the information indicating the layout of the document in the extracted character string and converts the extracted character string into individual extracted character strings. Be composed.

According to this dictionary creation device, if the extracted character string includes information indicating the layout of the document, the extracted character string can be divided based on the layout information and registered as individual headwords in the dictionary information.

Therefore, a headword is not registered in the dictionary information while layout information such as a line feed code is included, and the accuracy of the dictionary information is increased.

In the above-described dictionary creating apparatus, the adjustment processing unit includes a word number processing unit that suppresses registration in the extraction information of the extracted character string when the number of words in the extracted character string is equal to or less than a predetermined number. Is done.

According to this dictionary creation device, it is possible to prevent an extracted character string having a number of words equal to or less than a predetermined number from being registered in the dictionary information as a headword.

Therefore, a word whose extracted character string is registered in an existing dictionary of the first type of language, such as a verb,
In the case where the word is composed of one word such as an adjective or a pronoun, the basic word is registered in the dictionary information as a headword, thereby avoiding an incorrect translation result using the dictionary information.

In the above dictionary creating apparatus, when the length of the extracted character string including a predetermined number of words or less is less than or equal to the predetermined length, the adjustment processing unit sets the extracted character string of the extracted information in the extraction information. It is configured to include a character string length processing unit that suppresses registration.

According to this dictionary creation device, when the length of the extracted character string including a predetermined number of words or less is equal to or less than the predetermined length, the extracted character string is registered as a headword in the dictionary information. Never.

Therefore, even if the extracted character string is not longer than a predetermined length, even if the extracted character string is not longer than a predetermined number of words, for example, if the extracted character string is composed of a long character string word with a suffix or prefix added thereto, In, the extracted character string is registered in the dictionary information as a headword. On the other hand, when the length is shorter than the predetermined length, the word is not registered as a headword in the dictionary information because it is highly likely that the word is a basic word in the first type of language. Therefore, the accuracy of the headword registered in the dictionary information and the dictionary information itself are improved.

In the above-described dictionary creation device, the adjustment processing unit performs the processing for the extracted character string including the predetermined number of words or less.
When uppercase and lowercase characters are mixed, the system includes a case processing unit for maintaining registration of the extracted character string in the extracted information.

According to this dictionary creation device, even if the extracted character string is composed of words of a predetermined number or less, if the uppercase and lowercase characters are mixed, the extracted character string is registered in the dictionary information as a headword. Process as follows.

Therefore, even if the extracted character string is composed of words equal to or less than a predetermined number, if uppercase and lowercase characters are mixed, the extracted character string should be translated using a basic dictionary such as proper nouns. Is registered as a headword in the dictionary information as a non-word, so that the accuracy of the dictionary information is improved.

In the dictionary creation device, the adjustment processing unit determines the part of speech of the extracted character string based on the character string in the document subsequent to the extracted character string, and extracts the part of speech according to whether or not the part of speech corresponds to the predetermined part of speech. It is configured to include a part-of-speech processing unit that determines whether to delete the registration of the extracted character string in the information.

According to this dictionary creation device, the part-of-speech processing unit processes the extracted character string corresponding to the predetermined part of speech of the first type language so as not to be registered as a headword in the dictionary information.

Therefore, in the document, there is no entry in the dictionary information that directly indicates a word of the first type language, such as a verb or the like, which is registered in an existing dictionary. The accuracy of the information is improved.

In the above-described dictionary creating apparatus, an auxiliary data creating section for creating auxiliary data used to assist in creating dictionary information is further provided, and the adjustment processing section is created by the auxiliary data creating section. The adjustment processing is performed with reference to the obtained auxiliary data.

According to this dictionary creation device, the adjustment processing section performs adjustment processing with reference to the auxiliary data. Therefore, the accuracy of the headword registered in the dictionary information is improved.

In the above-described dictionary creation device, the auxiliary data creation unit includes a frequency information creation unit that counts the number of times that the extracted character string is extracted from the document and creates frequency information. The adjustment processing unit is configured to include a frequency information processing unit that deletes an extracted character string whose number of extractions is determined to be less than a predetermined number of times based on the frequency information created by the frequency information creation unit from the registration of the extracted information. Is done.

According to this dictionary creation device, the frequency information creation unit creates frequency information indicating the number of times each character string is extracted from the document, and the frequency information processing unit determines the number of extractions based on the frequency information. The extracted character string determined to be less than the predetermined number is not registered in the dictionary information as a headword.

Therefore, it is possible to determine whether or not to register an extracted character string in the dictionary information as a headword in accordance with the number of times that the character string appears in the document. Since it is possible to process an extracted character string that appears by accident because it has been forgotten to use it as a headword, the accuracy of the dictionary information is improved.

In the above-described dictionary creation device, the auxiliary data creation unit includes a dictionary lookup result creation unit that creates a search result by searching for an extracted character string from a dictionary group including dictionary information and an existing dictionary. A dictionary lookup processing unit that suppresses registration of the extracted character string in the extracted information when the search result created by the dictionary lookup result creation unit indicates that the extracted character string is already registered in the existing dictionary group It is comprised so that it may have.

According to this dictionary creation device, when the search result created by the dictionary lookup result creation unit indicates that the extracted character string is already registered in the existing dictionary group, the dictionary lookup processing unit performs the extraction. It is avoided that the information is registered in the dictionary information as a headword.

Therefore, in the above-described dictionary creating apparatus, the extracted information which has been registered in the existing dictionary group is registered in the dictionary information as a headword, and the capacity of the dictionary information can be suppressed from becoming unnecessarily large. The auxiliary data creation unit is configured to include a translation result creation unit that creates a translation result by translating an extracted character string with reference to a dictionary group.

According to this dictionary creating apparatus, the translation result using the dictionary group of the extracted character strings is used as data to assist in creating the dictionary information. Can be done.

In the above-described dictionary creation device, the adjustment processing unit includes the first translation processing unit that maintains the registration of the extracted character string in the extraction information when the translation result created by the translation result creation unit is not appropriate. It is configured to comprise.

According to this dictionary creation device, if the result of translating the extracted character string with reference to the dictionary group is not appropriate,
The extracted information is registered in the dictionary information as a headword.

Therefore, an extracted character string whose translation result is not appropriate, in other words, if the translation result is not proper when translated using the basic dictionary information, the extracted character string is included in the dictionary information as a headword. Since it is registered, at the time of subsequent translation, the translation result of the extracted character string is appropriate by referring to the dictionary information. In the above-described dictionary creation device, the adjustment processing unit, when the translation result created by the translation result creation unit indicates the extracted character string, suppresses registration of the extracted character string in the extraction information. It is comprised so that a part may be provided.

According to this dictionary creation device, if the result of translating the extracted character string with reference to the dictionary group indicates the extracted character string, the extracted information is registered in the dictionary information as a headword. Is suppressed.

Therefore, if the translation result is the extracted character string, in other words, if a certain extracted character string is translated using a dictionary group, and if the extracted character string is obtained as a translation result, This indicates that the extracted character string itself is already registered in the dictionary group. Therefore, in such a case, registration of the extracted character string in the dictionary information as a headword is suppressed, and duplicate registration of the extracted character string as a headword is avoided. Therefore, it is possible to maintain the dictionary capacity at an appropriate level and increase the dictionary search efficiency.

In the above-described dictionary creating apparatus, the dictionary information creating section is provided with a dictionary information correcting section for correcting dictionary information.

According to this dictionary creating apparatus, the created dictionary information is corrected by using the dictionary information correcting section. Therefore, the dictionary information can be corrected so as to have more appropriate contents.

In the dictionary creating apparatus described above, the headword information includes a translation of the corresponding headword, and the dictionary information correction unit converts the corresponding headword character into a full-width character as a translation and registers it. It is comprised including a conversion part.

According to this dictionary creation device, the translated word double-width unit of the dictionary information correction unit registers, in the translated word of the dictionary information, the translated word of the corresponding headword into full-width characters. Therefore, by referring to the dictionary information at the time of translation, it is possible to assign a word corresponding to the headword as a translation word in which the headword is converted to a double-width character of the first type language.

In the above-described dictionary creation apparatus, the headword information includes the translation of the corresponding headword, and the dictionary information correction unit includes a translation word determination unit that determines and adds a translation of the headword from the document. You.

According to this dictionary creation device, the translated word determined by the translated word determining unit from the contents of the document is added to the dictionary information in correspondence with the headword. Therefore, when the translated word of the headword is described in the document, it can be registered in the dictionary information as a corresponding translated word, so that the accuracy of the dictionary information is improved.

In the above-described dictionary creation device, the headword information includes the translation of the corresponding headword, and the dictionary information correction unit determines whether the character string is similar in one or more headwords in the dictionary information. Is configured to include a translated word information unifying unit for assigning the same translated word to the one or more headwords.

According to this dictionary creation apparatus, if the notation of the character strings in one or more headwords in the dictionary information is similar to each other by the translated word information unifying unit of the dictionary information correcting unit, The same translation is given to the above headword.

Therefore, even though the headword is a headword representing the same thing, it is registered in the dictionary information as another headword due to a difference of a small letter, a space, and the presence or absence of a hyphen symbol. Is suppressed from becoming larger than necessary.

In the dictionary creating apparatus described above, the information modifying unit for modifying the contents of the dictionary information created by the dictionary information creating unit as desired in accordance with information input from the outside via the input unit is provided. Further provided and configured.

According to the dictionary creating apparatus, the created dictionary information can be changed as desired by the user according to the information given from the outside by the information correcting unit, and the dictionary information customized by the user can be easily obtained. Or highly accurate dictionary information.

In the above-described dictionary creation device, the information correction unit includes an external designation unit for designating the contents of the adjustment processing by the adjustment processing unit in accordance with an external instruction via the input unit.

According to this dictionary creation device, the contents of the adjustment processing by the adjustment processing unit are specified by the external specification unit of the information correction unit in accordance with an external instruction via the input unit. Therefore, the user can variably adjust the adjustment processing content of the adjustment processing unit as desired by the external designation unit only by giving an instruction from the outside via the input unit, and can further customize the dictionary information.

In the above-described dictionary creating apparatus, the external designation unit variably designates whether or not one or more of the above-mentioned processing units of the adjustment processing unit is applied according to an external instruction via the input unit. It is composed of

According to this dictionary creation device, the external designation section variably designates whether or not to apply one or more of the above-described processing sections of the adjustment processing section in accordance with an external instruction through the input section. The user can externally designate a processing unit desired to be applied via the input unit, and obtain dictionary information created using the desired processing unit.

In the above-described dictionary creating apparatus, the information correction unit finally registers, in each of one or more processing units of the adjustment processing unit, a headword in the dictionary information to which the processing unit is applied, in the dictionary information. It further includes a registration determining unit for determining whether or not to perform through the input unit.

According to this dictionary creation device, the registration deciding unit makes each of at least one processing unit of the adjustment processing unit a
It is determined via the input unit whether or not the headword in the dictionary information to which the processing unit is applied is finally registered in the dictionary information. Therefore, the user can operate the input unit and arbitrarily select whether to maintain the registration in the dictionary information or to delete the registration of the headword adjusted by applying each processing unit by the registration determining unit. Thus, dictionary information customized by the user can be obtained.

The dictionary creating apparatus described above is configured so that the data created by the auxiliary data creating unit is presented at the time of designation by the external designation unit.

According to this dictionary creation device, at the time of designation by the external designation unit, the data created by the auxiliary data creation unit is presented. Therefore, the user can refer to the presented creation data, give a more accurate instruction regarding the adjustment content of the adjustment processing unit via the input unit, and obtain user-customized dictionary information.

A translation device according to still another aspect of the present invention performs a translation process from a language represented by a first type of character to a language represented by a second type of character. Is a character string composed of characters of the first type extracted from a document in which characters are mixed, and one or more headwords not registered in the existing dictionary, and information on each headword for each headword is provided as headword information. It has a dictionary storage unit in which the dictionary information stored as is stored, and is configured to refer to the information in the dictionary storage unit for translation processing.

According to this translator, even in a sentence of the second type language, a term which is usually described in the first type language and which is not registered in the existing dictionary (new words, technical terms) ) Is referred to when the document of the first language indicated by the first type of character is translated into the document of the second language indicated by the second type of character with reference to the dictionary information summarizing the above. Therefore, highly accurate translation processing can be performed.

In the above-described translation apparatus, the dictionary information is configured to be created by the above-described dictionary creation apparatus.

According to still another aspect of the present invention, there is provided a dictionary creating method for converting a first language represented by a first type of character into a second language.
A method for creating dictionary information in which a headword that is used for translation and used for translation and that is not registered in the existing dictionary information is registered using a plurality of types of characters. An input step for inputting information including a document containing mixed types of characters, and extracting one or more character strings of the first type of characters from the document input from the input step, and registering the extracted character strings A character string extracting step of outputting the extracted information, and an extracted character string of the extracted information output from the character string extracting step as a headword, and information relating to the headword corresponding to each of the headwords. And a dictionary information creating step of creating dictionary information by giving

According to this dictionary creation method, a character string composed of a first type of character is extracted from a document in which a plurality of types of characters are mixed, and a character string that is not registered in an existing dictionary is extracted and registered as a headword. The generated dictionary information is created, and the dictionary information referred to for translating the first language document represented by the first type of character into the second language document represented by the second type of character is generated. .

Therefore, even in the text of the second type language, it is a term (new word, technical term) that is daringly described in the first type language, and an unregistered headword is collected in the existing dictionary. By creating dictionary information, a document in a first language represented by a first type of character is converted to a second document represented by a second type of character.
By referring to the dictionary information when translating into a document in a language, translation processing with higher accuracy can be performed.

In the above-described dictionary creation method, an information correction step for correcting the contents of the dictionary information created in the dictionary information creation step as desired in accordance with information input from the outside via the input step is provided. In addition, it is configured.

According to such a dictionary creation method, the created dictionary information can be changed as desired by the user according to information given from the outside, for example, by the information correction step, and can be easily customized by the user. Highly accurate dictionary information can be obtained. By referring to the dictionary information, more accurate translation can be performed.

A recording medium according to still another aspect of the present invention stores a dictionary creating program for causing a computer to execute a dictionary creating method for creating dictionary information used for translating from a first language to a second language. This is a computer-readable recording medium having the following characteristics.

In other words, the dictionary creation method is based on the first language represented by the first type of character and the second language represented by the second type of character.
An existing dictionary used for translating to a language and used for translation is a dictionary creation method for creating dictionary information in which unregistered headwords are registered, and includes a document in which a plurality of types of characters are mixed. An input step for inputting information, and a character string for extracting one or more character strings of a first type of character from a document input from the input step, and outputting extracted information in which the extracted character strings are registered An extraction step;
A dictionary information creating step of creating dictionary information by using the extracted character string of the extracted information output from the character string extracting step as a headword and adding information about the headword as headword information corresponding to each of the headwords; And

According to the dictionary creation method executed by the program recorded on the recording medium, a character string composed of the first type of character is extracted from a document in which a plurality of types of characters are mixed and registered as a headword. Dictionary information is created,
Dictionary information referred to for translating a first language document represented by a first type of character into a second language document represented by a second type of character is created.

Therefore, even in the text of the second type language, it is a term (new word, technical term) usually described in the first type language, and the unregistered entry words are collected in the existing dictionary. The dictionary information can be created so as to refer to the dictionary information when translating a first language document represented by a first type of character into a second language document represented by a second type of character. If this is the case, more accurate translation processing can be performed.

The dictionary creation method executed by the above-mentioned program further modifies the contents of the dictionary information created in the dictionary information creation step as desired according to information input from the outside through the input step. And an information correction step for the information processing.

According to this dictionary creation method, the created dictionary information can be changed as desired by the user in accordance with information given from the outside, for example, by the information correction step, and can be easily customized by the user to achieve high accuracy. High dictionary information can be obtained.

[0089]

Embodiments of the present invention will be described below.

FIG. 1 is a diagram showing a functional configuration of a dictionary creation device according to an embodiment of the present invention. FIG. 2 is a diagram showing a hardware configuration of the dictionary creation device according to the embodiment of the present invention. FIG. 3 is a flowchart showing a schematic processing procedure for creating a dictionary using each function of FIG.

Referring to FIG. 2, dictionary creating apparatus 500 includes:
Various data including programs between the input device 100, the output device 200, the storage unit 300, a central processing unit (CPU) 400 for centrally controlling and managing the device itself, and an external storage medium 33. And a medium access device 34 for transferring the data.

The input device 100 includes a keyboard, a mouse,
It includes a device for externally operable data input such as a pen / tablet and a communication device for inputting data via an external communication network.

The output device 200 includes a display device including a liquid crystal display and a plasma display, and a printer including a laser printer and a thermal printer.

ROM (Read Only Memory) and RA
The storage unit 300 including M (random access memory) includes a data storage unit 3 and a program storage unit 301
And In the program storage unit 301, processing programs in each functional unit of FIG. 1 shown in each flowchart described later are stored in advance and executed under the control of the CPU 400.

FIG. 4 is a diagram showing an example of the contents stored in the data storage section 3 of FIG. As shown in FIG. 4, the data storage unit 3 stores the dictionary data 30 created by the dictionary creation device 500, the document data 40 input via the input device 100 to create the dictionary data 30, and various types of data in the dictionary creation. Rule data 50 in which rules are registered, basic dictionary data 60 which is data of a dictionary which is generally used as a basis,
The specialized dictionary data 70, which is dictionary data applied to a specialized field, and various data 80, which is data used in various processes in the device 500, are registered.

The dictionary data 30 includes a plurality of different headwords 3
1 includes the dictionary information 32. The dictionary information 32 is a translation 3 obtained by translating the corresponding headword 31.
2A and the corresponding part of speech 32B.

The rule data 50 includes one or more rule types 51 applicable to the creation of the dictionary data 30.
For each of the rule types 51, there is included application presence / absence information 52 for designating whether or not the rule is applied when the dictionary data 30 is created.

Next, with reference to FIG.
Will be described. In FIG. 1, a dictionary creation device 500 has an input unit 1 for inputting document data 40 and also various designation information of a user, an output including display of various data including the created dictionary data 30, or an external storage medium. An output unit 2, a data storage unit 3, a character string extraction unit 4, a headword recognition unit 5, a dictionary information creation unit 17, a dictionary information correction unit 18,
It includes a user support unit 22 and an auxiliary data creation unit 29.

Each of the input unit 1, headword recognition unit 5, dictionary information correction unit 18 and user support unit 22 is included in the input unit, adjustment processing unit, dictionary information correction unit and information correction unit of the present invention. .

The character string extracting unit 4 extracts a specific type of character string from the document data 40 input from the input unit 1. The headword recognizing unit 5 adjusts the character string extracting unit 4 so that an inappropriate character string is not extracted. For this purpose, the headword recognition unit 5 processes a character string before and after a parenthesis symbol, a parenthesis symbol removal / separation unit 6, a specific symbol division unit 7 that divides a character string before and after a specific symbol, and removes a numeric string in the character string. A string information removing unit 8 for dividing or extracting a character string using the layout information, and a character string having a number of words less than or equal to a specified number by counting the words in the character string. , A character string length check unit 11 for detecting the length of a character string and preventing a character string having a length less than a prescribed length from being extracted,
A lower case checker 12 for determining lower case, a part of speech checker 13 for determining the part of speech of the extracted character string by looking behind the extracted character string, and a frequency information generator 290 described later based on frequency information created by a frequency information generator 290. In addition, based on a dictionary lookup result created by a dictionary lookup result creation section 291 to be described later, a frequency information check section 14 for extracting only a character string that appears a certain number of times or more in the document 40, and a registered word Dictionary lookup check unit 1 to check that data is not additionally registered in the dictionary
5. The translation checking unit 16 includes a translation checking unit 16 that checks, based on a translation result created by a translation result creation unit 292 described later, that words having the same translation result are not registered in the dictionary redundantly.

The dictionary information creating unit 17 assigns the dictionary information 32 including the translated word 32A and the part of speech 32B to the headword 31 which is the character string extracted by the character string extracting unit 4.

The dictionary information correction unit 18 is provided with the dictionary information creation unit 1
In order to correct the contents of the dictionary data 30 created by the step 7, the translation word full-width unit 19, the translation word information extraction unit 20, and the translation word information unification unit 21 are included. The translated word double-byte unit 19 converts the extracted character string into a double-byte character string and registers it in the dictionary data 30 as a headword 31. The translated word information extraction unit 20 estimates and assigns the translated word 32A from the input document 40 to the headword 31. When there is at least one similar notation of the extracted character string, the translated word information unifying unit 21 assigns a unified translated word 32B to each of the similar notations.

The user support unit 22 performs a support process to edit the contents of the dictionary data 30 as desired by the user. For this purpose, the dictionary data 30
Dictionary information correcting unit 23 for correcting the contents of the dictionary, an application rule indicating unit 24 for selecting a rule specified by the user, and setting various constants relating to the process of creating a dictionary, and registering the user in the dictionary data 30. Rule instructing unit 25 that can select a rule for assigning a registration pending mark to make it easier to determine the presence or absence of a frequency of acquiring frequency information from a frequency information creating unit 290, which will be described later, as a reference for determining user instructions. An information creation instruction unit 26, a dictionary lookup instruction unit 27 that acquires a dictionary lookup result from a dictionary lookup result creation unit 291 also described later, and a translation result creation unit 292 also described later.
And a translation instructing unit 28 for acquiring a translation result from the.

The dictionary information correction unit 23 and the application rule designating unit 24 are included in the dictionary information desired correction unit and the external designation unit of the present invention.

The auxiliary data creation unit 29 creates data for assisting the processing in the headword recognition unit 5 and the user support unit 22. The auxiliary data creation unit 29 stores the document data 4
A frequency information creation unit 290 that counts how many times the same character string is extracted from 0 by the character string extraction unit 4, a dictionary lookup that searches the character string extracted from the document data 40 from the headword 31 of the existing dictionary data 30 It includes a result creation unit 291 and a translation result creation unit 292 that translates the extracted character string using various dictionaries in the data storage unit 3.

Next, with reference to the processing flowchart of FIG. 3, the dictionary creation processing operation of the dictionary creation device 500 will be described with reference to the configuration of FIG. Here, it is assumed that an English character string is extracted from the input document data 40, which is a Japanese document, and registered in the dictionary data 30 as a headword 31.

FIG. 5 is a diagram showing an example of the input document data 40 of FIG. First, when the user inputs the document data 40 via the input unit 1 (document input in S1 of FIG. 3), S
The second and subsequent processes are started.

On the other hand, when an instruction to modify the dictionary data 30 to modify the contents of the created dictionary data 30 as desired by the user is input, the contents of the dictionary data 30 It is modified as desired (S10). The details will be described later.

Now, when the user inputs the Japanese document data 40 as shown in FIG. 5 through the input unit 1, it is stored in the data storage unit 3 and given to the character string extraction unit 4 for processing ( S2).

For example, the input document data 40 in FIG. 5 may be a document created by a human or a translation result of an English-Japanese machine translation device. However, HTML (Hyper Text Mark-up L
In the case of a text formatted like anguage or RTF (Rich Text Format), it is assumed that the document has tags and decorative character information removed (unnecessary information such as tags and decorative character information). (It is easy to remove (out-of-language information) by using existing technology, so the description is omitted.)

Referring to FIG. 3, character string extracting section 4 extracts a specific type of character string from input document data 40. At this time, the headword recognizing unit 5 performs an adjustment process so that an inappropriate character string is not extracted (S3). Also, a headword recognition unit 5
The data referred to to assist the adjustment process by the auxiliary data generator 29 is generated by the auxiliary data generator 29 (S4).

The character string extracted by the character string extracting unit 4 is converted into the dictionary data 30 by the dictionary information creating unit 17 into the dictionary data 30.
And the corresponding dictionary information 32 is created and registered in the dictionary data 30 (S5). At this time, the created dictionary information 32 is corrected and registered by the dictionary information correction unit 18 (S6).

Thereafter, the dictionary data 30 is read from the data storage unit 3 by the output unit 2 and output to the output device 200 or the external storage medium 3 via the medium access device 34.
3 and registered.

FIG. 6 is a processing flowchart of the character string extracting unit 4 of FIG. FIG. 7 is a diagram showing the extracted information 44 obtained by the processing of FIG. The extraction information 44 is stored in the data storage unit 3 as various data 80.

As described above, the Japanese document data 40 read by the input unit 1 is first passed to the character string extraction unit 4. The character string extraction unit 4 stores the headword 31 in the dictionary data 30.
A character string to be registered as is extracted. Since the document data 40 is a Japanese document, the document data 40 is mostly composed of Japanese character strings, but includes “Copernicus” on the fourth line and “Zaur
As shown in “us” and “ReadMe Firs” in the seventh line, an English character string appears. Since these English character strings express proper nouns such as product names and person names or technical terms, and have no translated word corresponding to Japanese or a well-known name or term, the author of the Japanese document data 40 The character string in the original language (English) is used as it is.

The dictionary creating apparatus 500 according to the present embodiment
Extracts a character string (word string) written in such another language / character from the document data 40 and performs processing to register it as a headword 31 of the dictionary data 30.

Here, a method of extracting a character string by the character string extracting unit 4 will be described with reference to the processing flow of FIG.

First, the variable p indicating the position of the character of the input document data 40 is set to p = 0 and read from the head (p = 0) of the document data 40 (S201 to S20).
2), detecting the location where the language type is switched (S204)
To S206). In the document data 40, the part where the characters are switched from Japanese characters to English characters is "Copernicus".

Here, the procedure for detecting the place where the Japanese character is switched to the English character is as follows. Unlike English characters (alphabet, numbers, and symbols) that can be expressed in one byte (8 bits) (half-width characters), Japanese characters (kanji, hiragana, katakana) have a very large number of characters Therefore, each character is represented by 2 bytes (16 bits) (full-width character). However, in order to distinguish between 2 English characters (2 bytes) and 1 Japanese character (2 bytes), 2 Japanese 1 characters
Several systems (Kanji code system) for using bytes
There is. In JIS, the front and back of a Japanese character string is sandwiched between specific codes, and in shift JIS, the characters are shifted to a specific area,
With the UC (abbreviation for Extended Unix Code) code, the most significant bit is set to identify it as two English characters (two bytes).

In this embodiment, the difference between the full-width character expression of Japanese characters and the half-width character expression of English characters is used.
A switching point between Japanese characters and English characters in the Japanese document data 40 is detected. For example, in the document data 40, the half-width start position S of the location where "...-
(S207).

Similarly, the position where the English character returns to the Japanese character is stored as the half-width end position E (S208). A character string from the half-width start position S to the half-width end position E is extracted and stored as the extracted character string 4A in the extraction information 44 of FIG. 7 (S209).

Thereafter, the same processing is performed, and until reading of the last character of the document data 40 is completed until there are no more characters to be input (S203), the Japanese document data 40 is read.
All of the English character strings therein are extracted as extracted character strings 4A and stored in the extraction information 44. In addition, since the same English character string may appear many times in the same document data 40, an appearance frequency 4B indicating the number of appearances is also registered for each extracted character string 4A. The extraction position of the extracted character string 4A in the document data 40 is registered using the positions S and E.

The processing of the headword recognition unit 5 (S21)
0) will be described later. In the present embodiment, the extraction of the English character string from the Japanese document data 40 has been described as an example, but the present invention is not limited to this. An English character string appearing in the document data 40 written in a language other than Japanese (for example, Korean) can be used as the extracted character string 4A.
It is also possible to extract a Japanese character string from the English document data 40 as the extracted character string 4A. Further, a kanji phrase, a hiragana character, a katakana character string in the Japanese document data 40 can be extracted as the extraction character string 4A.

In the present embodiment, the description has been made based on the character switching position information. However, for languages composed of almost the same character type, such as English and German, for example, German A word including a character unique to one language, such as a word umlaut symbol, may be extracted as the extracted character string 4A. Further, such a difference between British English and American English having no difference in character type can be made by checking the basic dictionary data 60 corresponding to each language.

FIG. 8 is a processing flowchart of the dictionary information creating unit 17 of FIG. FIG. 9 is a diagram showing an example of the dictionary data 30 obtained by the processing of FIG. The dictionary information creation unit 17 refers to the extraction information 44 in FIG. 7 obtained by the character string extraction unit 4 and first determines whether there is an extracted character string 4A in order from the top, and if it is lost (NO in S500), processing Is terminated, but if there is an extracted character string 4A, the data storage unit 3
Each of the extracted character strings 4A is registered as a headword 31 in the dictionary data 30 of FIG. 9 and the corresponding dictionary information 32 is registered to obtain the dictionary data 30 of FIG. 9 (S500 and S500).
501 loop processing).

In this case, a noun (noun) is set in the part of speech 32B in order to translate an English word (English character string) for which the translation is not yet determined, and the translation 32A has the same character as the corresponding headword 31. The column is set.

In the same manner, all English extracted character strings 4A extracted from the Japanese document data 40 are registered as headwords 31, and dictionary information 32 including corresponding parts of speech 32B and translated words 32A is registered. Thus, dictionary data 30 is formed. As a result, the document data 40
From this, the dictionary data shown in FIG. 9 is obtained.

FIG. 10 is a block diagram of a translation apparatus that uses the dictionary data 30 created by the dictionary creation apparatus of FIG. 1 as a part of dictionary data referred for translation.

The dictionary data 30 created in FIG.
The data is read from the data storage unit 3 under the control of U400 and recorded in the external storage device 33 via the medium access device 34.

Referring to FIG. 10, translation apparatus 501 has a CPU
A data access unit 341 for accessing the data storage unit 700 for translation, an output unit 800, an input unit 900, and the external storage medium 33. The dictionary data 30 created by the dictionary creation device 500 of FIG. 1 and recorded in the external storage medium 33 is read by the medium access device 341 of the translation device 501, and is translated via the CPU 600 into the data storage unit 7 for translation.
The translation dictionary data 710 is registered in the translation dictionary data 710.

The translation device 501 regards the dictionary data 30 created as described above as a set of English words (non-translated words) not desired to be translated into Japanese and translates them into the translation dictionary data 710 as a user dictionary for English-Japanese translation. If you register and use it, it is a new term such as a proper noun or technical term such as a product or person, and there is no translated word corresponding to Japanese, it is not fixed and it is forcibly translated into Japanese Thus, a translation result that is easier to read when translated as an English word can be obtained.

For example, the following sentence “This Take Out is Popula” including the name of some product “Take Out”
".r." as it is, the translation result is broken as "this goes out, is popular." Therefore, if the dictionary data 30 created by the dictionary creation device 500 is registered as the translation dictionary data 710, , "Take Out is popular," you get the appropriate translation.

FIGS. 11 to 13A and 13B are processing flowcharts of the headword recognizing unit 5 of FIG. FIG.
1 shows the processing of the parenthesis symbol removal / separation unit 6, the specific symbol division unit 7, and the numeric string removal unit 8 of the headword recognition unit 5, and FIG.
Reference numeral 2 denotes a layout information check unit 9, a word count check unit 10, a character string length check unit 11, and a case check unit 1.
2. Part of speech checker 13 and frequency information checker 14
13A and 13B show the processing of the dictionary check unit 15 and the translation check unit 16.

The above-described character string extracting section 4 simply extracts a character string of only the information of the location where the language has been switched, and registers the extracted character string in the dictionary data 30 as a headword 31. Therefore, according to this registration method, for example, when there is a character string “Here, machine translation (Machine Translation) is described”, the parentheses are half-width, so “Machine Translat
ion "is extracted and registered as the entry word 31.

In order to exclude such an inappropriate entry 31 in advance, in the present embodiment, an appropriate entry 31 is selected from among the character strings 4A extracted by the above-described character string extraction unit 4. In order to recognize only the extracted character string 4A, a headword recognition unit 5 that operates every time the character string 4A is extracted is provided. The processing of the headword recognition unit 5 is described in FIGS.
3A and 3B will be described.

6: Bracket symbol removal / separation unit First, the English character string 4A extracted like “Machine Translation has an old history” → extracted character string “(Machine Translation)” Is included, the parenthesized symbol removing / separating unit 6 is applied, and FIG.
S301 to S307 are executed.

In order to apply the parenthesis symbol removal / separation unit 6, a parenthesis symbol position 4D (a pair of opening and closing if possible) is registered in the character string extraction unit 4 as shown in FIG. . The processing in the parenthesis symbol removal / separation unit 6 depends on the appearance pattern of the parentheses. Specifically, "Electronic Newspaper (Elec
tronic Newspaper…) → “(Electronic Newspape
If the beginning or end of the English character string 4A extracted as “r)” is a parenthesis symbol (S302), the character string “Electronic Newspaper” is extracted after removing the parentheses according to the parenthesis symbol position 40. The character string 4A is registered in the extraction information 44.

In addition, "... and digital electronic library DL
(“Digital Library”) → “DL (Digital Lib
rary) ”, and there is an additional English character string (DL) before the open parenthesis symbol (S304).
Character string "Digital Libr" in parentheses according to parenthesis symbol position 4D
In addition to extracting “ary” as the extracted character string 4A, the leading English character string “DL” representing the abbreviation is registered in the extraction information 44 as the extracted character string 4A so as to be another headword (S305).

[... (C) Sharp Corporation ...] →
The character string “(C) Sharp Corporation” is registered in the extraction information 44 as it is as the extracted character string 4A. As described above, when the entire parentheses are the symbol (C) representing copyright and the symbol (R) representing trademark registration (S306), the entire parentheses are deleted and only the character string "Sharp Corporation" is extracted as the extracted character string 4A. Register in information 44.

7: Specific symbol division unit Next, for example, "High-speed 100BASE-TX / 10B
Equipped with ASE-T (RJ45 connector) LAN ... "→
"100BASE-TX / 10B" is used as the extracted character string 4A.
When ASE-T "is obtained, and when a slash symbol (" / ") representing a parallel phrase is included (S308), the specific symbol division unit 7 is applied. Therefore, it is necessary that the character string extraction unit 4 recognizes a symbol ("/") to be divided in advance and stores it as various data 80.

In the present embodiment, words are separated before and after “/”, and then extracted separately as extracted character strings 4A.
(S309). In the character string described above, “10
Two extracted character strings 4A of “0BASE-TX” and “10BASE-T” are obtained as the extraction information 44. Although “/” is mentioned here, for example, division of “:”, “;”, “-”, etc. Symbols can be applied in a similar manner.

8: Numeric string remover Next, the character string "Power E / J3.0 is replaced with Power E / JV2.
Compared to 1 ... "→ Extracted character string 4A as" Power E / J
When a digit string is included at the end, such as “3.0” or “Power E / JV2.1” (S310), the digit string removing unit is applied. The above-described character string extracting unit 4 needs to recognize the numeral string to which the numeral string removing unit 8 is applied and the related character strings (., Ver, V, etc.) and register them in advance as various data 80.

When the extracted English extracted character string 4A is a number string at the end, for example, a number string meaning version information (may include a decimal point) (S310),
If this number string is extracted as the extracted character string 4A, it will not be possible to look up the dictionary as a headword 31 only by changing the version number. By registering with the extracted character string 4A from which the number string has been removed (S311), the versatility of the headword 31 in the dictionary data 30 is improved, and the capacity of the dictionary data 30 can be reduced.

In the above example, “Power E / J3.0” is not registered as the extracted character string 4A for both “Power E / JV2.1”.
r E / J "only. In the latter case, in particular," Ver. "
"V.", "Ver."
Numerals and the like can also be removed by making them the targets of this processing.

9: Layout information check section Next, the character string "@Home Page:-Released MD Studio (MD-F20) ...-Released liquid crystal monitor (LL-T142A) ..."-> Character string as extracted character string 4A Home Page: <Line break>
The layout information check unit 9 applied when the extracted character string 4A is an English character string and includes a line feed code or the like as in “−MD” will be described.

In the layout information check section 9, the extracted character string 4A is synthesized with a character string of a different line including layout information such as a character string "Home Page: <line feed>-MD" which is non-language information such as a line feed code. If the extracted character string 4A is recognized (S312), the extracted character string 4A is separated at the position of the line feed code and registered in the extraction information 44 as an individual extracted character string 4A (S313).

Here, the line feed code is given as the layout information, but there are also information on paragraphs (blank lines or indentation information), information on bullet points, and information for modifying character strings such as fonts and underlines. These pieces of information can be used as information for separating the same character string.

Note that these pieces of layout information are separated when the document data 40 input by the input unit 1 is read, but need to be saved when the layout information check unit 9 is applied.

10: Word Number Checking Unit For example, in the character string “baseball game Play is...”, The character string extracting unit 4 extracts the English character string “Play” as the extracted character string 4A, but one English word Is satisfied, the word number check unit 10 is applied.

When the extracted character string 4A is one basic English word as described above, if it is registered as the headword 31 in the dictionary data 30 as it is, the original translation other than the product name (in this example, Not only is it not translated when used in "play"), but in most cases used in verbs, the part-of-speech is erroneously translated and the translation result is more likely to be broken.

Therefore, in the present embodiment, the word number checking unit 1
0 is applied to distinguish between words and idioms, and in the case of one word (S314), processing such as not registering the extracted character string 4A in the extraction information 44 is performed, and the above-described English word “Pla
Here, “y” is not registered. Here, the word count checking unit 10 determines that the word is a idiom if it consists of two or more words, but determines whether the word is a word based on whether there is a blank (tab) due to a character string. Is also good.

11: Character String Length Checking Unit Next, for example, the character string “newspaper distribution service NewspaperSel
In the case of “f is...”, the extracted character string 4A (“NewspaperSelf”) extracted by the character string extraction unit 4 is one word but long, so in this case, the character string length check unit 11 is applied. When the registration of the one word (“NewspaperSelf”) as the extracted character string 4A in the extraction information 44 by the application of the word number checking unit 10 described above is excluded, the translation using the dictionary data 30 created using this is performed. Depending on the unknown word processing of the device, it is also assumed that the dictionary is separated into Newspaper and Self, and the translations ("newspaper" and "self") are configured to generate word translations ("newspaper-self"). So a long English string that may contain a suffix is 1
It is desirable that even words are registered in the dictionary data 30 (S316).

12: Case Checking Unit Next, in the character string “Our search engine KeyWord is...”, The character string extracting unit 4 extracts the “KeyWor
If uppercase and lowercase letters are mixed as in “d” (S31
7), the case checker 12 is applied.

By applying the above-described word count checking unit 10, the registration of the above-mentioned extracted character string 4A “KeyWord” in the extraction information 44 is excluded. Since it is very likely that the word is a proper noun, it is desirable to register the word as an untranslated word in the extraction information 44 as the extracted character string 4A and to register it as the index word 31 in the dictionary data 30 (S318). Conversely, it is also useful to determine whether or not to register based on the size of the characters constituting the character string, such as not being registered as the extracted character string 4A and the headword 31 if all are lowercase.

13: Part-of-speech checking unit Next, in the case of the character string "this property is widely propagated.", The character string extracting unit 4 extracts the English character string "propagate" used as a verb as the extracted character string 4A. However, in this case, the part-of-speech checking unit 13 is applied.

In the above description, the case where the English character string extracted as the extracted character string 4A is a noun such as a product name has been mainly described, but the document data 40 is data quoted from the technical literature. In some cases, English words other than nouns may be used. If the extracted character string 4A is registered as the headword 31 in the dictionary data 30 and the noun is registered in the part of speech 32B, the translation result using the dictionary data 30 will fail. May be. However, as described above, when a Japanese verb ("propagate") is used as a part of a verb ("propagate"), the part of speech is identified by identifying the Japanese that follows the English extracted character string 4A extracted as follows. Can be determined, and incorrect dictionary registration can be avoided.

In other words, in the case of Japanese, when a non-verb word (noun or foreign word) is converted to a verb, it becomes a sa-varib. Therefore, the Japanese character string following the English character string as the extracted character string is
If any of the following is true, the extracted character string 4A is a verb, and thus is not registered as the extracted character string 4A in the extraction information 44 and is further registered as the headword 31 in the dictionary data 30. Can be avoided (S32
1).

Here, the auxiliary data generator 29 will be described. Each unit of the headword recognizing unit 5 described above can perform processing to avoid erroneous dictionary registration based only on the character string near the extracted character string 4A in the input document data 40. If it is possible to refer to the auxiliary data to be assisted, it becomes possible to recognize the extracted character string 4A that becomes the more appropriate headword 31. The auxiliary data creation unit 29 includes a frequency information creation unit 290, a dictionary lookup result creation unit 291, and a translation result creation unit 292 in order to create data for assisting the headword recognition process with reference to the headword recognition unit 5. Having. Hereinafter, the auxiliary data creation unit 29
The processing of the frequency information check unit 14, the dictionary lookup check unit 15, and the translation check unit 16 using is described.

The auxiliary data creation unit 29 is used not only by the headword recognition unit 5 but also by the user support unit 22 described later.

14: Frequency Information Checking Unit As a result of extracting the character string by the character string extracting unit 4 in the document data 40, the English extracted character string 4A that appears only once in the document data 40 is determined by the author of the document data 40. May have forgotten to use the translated word, or the extraction process of the character string extraction unit 4 may be incorrect (for example, two English character strings that happen to appear consecutively are recognized together). Conversely, it is certain that the extracted character string 4A that appears many times and has a high appearance frequency 4B can be used as a translated word as it is without any problem.

Therefore, the frequency information check unit 14 acquires the appearance frequency 4B corresponding to the extracted character string 4A recorded in the extraction information 44 by the character string extraction unit 4 from the frequency information creation unit 290, and appears only a predetermined number of times or less. Check that the extracted character string 4A that has not been entered is not registered as a headword 31 in the dictionary data 30. For example, the extracted character string 4A is extracted from the extraction information 44.
Remove.

For example, the character string "Home Page: <line feed>-MD" which is the extracted character string 4A connected to the preceding and succeeding lines used in the above-described layout information check unit 9 appears only once in the input document 40. However, since the value of the corresponding appearance frequency 4B is 1, the frequency information check unit 14 can exclude the frequency from the extracted information 44 (S323).

15: Dictionary Lookup Checking Unit As described above, the extraction of the character string 4A that becomes an inappropriate headword 31 that adversely affects the translation result using the dictionary data 30 is performed by the dictionary lookup checking unit 15 By using the check, the information can be more reliably excluded from the extracted information 44. For example, the dictionary lookup result creating unit 291 extracts the extracted character string 4A in the extracted information 44 from the dictionary data 3
0, the dictionary is searched using the basic dictionary data 60 and the specialized dictionary data 70 (S324), and according to the result, the following processing can be performed (S325).

That is, if the headword 31 has already been registered in the dictionary data 30 being created (Y in S325), the extracted character string 4A is not registered in the dictionary data 30 and the corresponding appearance frequency 4B in the extraction information 44 is not registered. Just count up. Also, a completely different dictionary, for example, the translation device 501
If it exists as a headword in the basic dictionary data 60 or the specialized dictionary data 70 applied to the
327), processing is performed so as not to be registered as the headword 31 in the dictionary data 30, but if the headword is not a basic headword (N in S327), the headword 31 is added to the dictionary data 30.
Process to be registered as If these dictionary lookups fail (N in S325), processing is performed so that the dictionary data 30 is newly registered as the headword 31 (S325).
326).

16: Translation Checking Unit Further, the extraction information 44 is checked by using the translation checking unit 16 so that an inappropriate entry word 31 that adversely affects the translation result is not registered in the dictionary data 30, thereby achieving higher accuracy. Dictionary data 30 with a high level FIG.
As shown in (A), words are supplemented before and after the extracted character string 4A to make a sentence, which is then translated by the translation result creation unit 292 (S329). If an incorrect translation result is obtained (N in S330), The extracted character string 4A is registered as it is in the extraction information 44 (S332).

The extracted character string 4A is, for example, a character string “Take
Out ”, a sentence“ This is Take Out. ”In which words are added before and after the character string is created, and the dictionary data 30, the basic dictionary data 60 and the specialized dictionary data 7
If the translation is made by the translation result creation unit 292 using 0, the translation result becomes “this is, it goes outside” (N in S330). In this case, the translation check unit 1
6 so that the “Take Out” of the extracted character string 4A is registered in the dictionary data 30 as the entry word
Is registered as an extracted character string 4A (S332). By the way, the extracted character string 4A is stored in the dictionary data 30 as the headword 3
When the above sentence is translated by using the dictionary data 30 after being registered as “1”, a translation result of “This is Take Out” can be obtained, which is effective.

In the translation result check section 16, the same headword is duplicated based on the translation result.
Processing is performed so as not to be registered in 0. As a result, it is possible to prevent the amount of registered data in the dictionary data 30 from unnecessarily increasing, and it is possible to reduce a dictionary search time during a translation process using the dictionary data 30. For its processing,
As shown in FIG. 13B, the translation check unit 16 supplements words before and after the extracted character string 4A to form a sentence, and translates the sentence using the translation result creation unit 292 (S329). If it indicates that it is the same as the character string 4A (Y in S330a), the extracted character string 4A
No. 4 (S331). If the same translation result is not obtained (N in S330a), the extracted character string 4A is registered in the extraction information 44 (S332).

Specifically, if the extracted character string 4A is, for example,
If the product name is the string "YourBestChoice"
The sentence "This is. YourBes" with words added before and after the character string
"tChoice" is created and translated by the translation result creating unit 292 using the dictionary data 30, the basic dictionary data 60, and the specialized dictionary data 70. At this time, a translation result "This is YourBestChoice" is obtained. If (Y in S330a), “Your” of the extracted character string 4A
Since “BestChoice” is already registered as a headword 31 in the dictionary data 30, the extracted character string 4A is not registered in the extraction information 44 in order to prevent the same headword 31 from being registered twice (S331). ).

FIG. 14 is a processing flowchart of the dictionary information correction unit 18 of FIG. Dictionary information creation unit 17 described above
In FIG. 9, the same character string as the headword 31 is given as the translated word 32A as shown in FIG. 9, but by using the dictionary information correction unit 18, other translated words 32A can be given. For this purpose, the dictionary information correction unit 18 includes a translated word full-width unit 19, a translated word information extraction unit 20, and a translated word information unifying unit 21. The operation of each of these units will be described below.

19: Translated Word Double-byte Unit In the dictionary data for English-Japanese machine translation, the translated word 32A is generally Japanese (double-byte characters). Therefore, the translated word full-width unit 19 double-widths the character string of the headword 31 as the translated word 32A and registers it in the dictionary data 31 (S400). FIG.
FIG. 15 is a diagram illustrating an example of a processing result of the translated word full-width unit 19 of FIG. 14.

20: Translated word information extraction unit An English character string sandwiched between parentheses, such as the character string "Electronic Newspaper" shown in the parenthesis symbol removal separation unit 6, is extracted as the extraction character string 4A. If so, the translated word information extraction unit 20 is applied.

As shown in the character string of this example, when the character string pattern has a double-byte character string before the opening parenthesis (S401), the double-byte character string immediately before the opening parenthesis is a translation of the English character string inside the parenthesis. Often it is. At this time, the character string “Electronic
“Newspaper” is registered in the extraction information 44 as the extracted character string 4A and is registered in the dictionary data 30 as the headword 31. In the present embodiment, a double-byte character string immediately before the parenthesis symbol (in this case, “electronic newspaper”) ) Is also extracted, and the dictionary data 3
0 (S402).

On the other hand, when the parentheses are half-width characters and the parentheses are full-width characters, for example, the character string “Electronic Newsp
aper (electronic newspaper) ”(Y in S401), the inside of the parentheses is registered in the dictionary data 30 as the translated word 32A (S40).
2). FIG. 16 is a diagram showing an example of the processing result of the translated word information extraction unit 20 of FIG.

21: Translation information unification unit Next, even though words that represent the same thing have subtle differences, such as differences in case and the presence or absence of a space or a hyphen symbol, another word (headword / translation word) is used. ) Is applied in order to avoid this in the case where it is registered as ()).

For example, the name Power E / J representing a product name may be different depending on the writer or even the same writer.
In some cases PowerE / J, PowerE / J, PowerEJ
In many cases, the writing style is slightly different. Therefore, in the present embodiment, similar headwords 31 differing only in the case, the presence or absence of a space or a hyphen are detected from the dictionary data 30 (S403), and the same is detected for all the detected similar headwords 31. The translated word information unifying unit 21 performs processing so that the translated word 32A is added. Thus, the created dictionary data 30 can also be used as a dictionary for term unification. At that time, the translated word 32A has the highest frequency of appearance 4
It is also possible to unify with the translated word 32A for the headword 31 corresponding to the extracted character string 4A having a high B (S404).

Next, the processing of the user support unit 22 will be described. FIG. 17 is a diagram showing an example of a display screen by the user support unit 22 of FIG. FIG. 18 is a diagram showing another example of the display screen by the user support unit 22 of FIG. FIG. 17 shows an example of a screen displayed on the output unit 2 in the processing of the dictionary information correction unit 23, the frequency information creation instruction unit 26, the dictionary lookup instruction unit 27, and the translation instruction unit 28 of the user support unit 22. FIG. 18 shows an example of a display screen at the time of the processing operation of the application rule instruction unit 24 and the hold rule instruction unit 25.

As described above, all of the headwords 31 and the dictionary information 32, which are information necessary for creating the dictionary data 30, are automatically generated from the document data 40 in which a plurality of languages to be extracted are mixed. I was However, inappropriate information may be registered in the dictionary data 30. Therefore, in the dictionary creation device 500, the user support unit 22 is applied so that the user can check the contents of the automatically created dictionary data 30 and correct the contents if necessary. User support unit 2 according to the flowchart of FIG.
The process 2 will be described.

In the state where the dictionary data 30 has been created and registered in the data storage unit 3 as described above, when an instruction to modify the dictionary data 30 is input through the input unit 1, the user support unit 22 The process is started (S10).

First, the contents of the dictionary data 30 created are displayed to the user via the output unit 2 (S11).
Then, it is asked whether the content is corrected for each headword (S
12). When the user indicates via the input unit 1 that the content is to be corrected for each headword 31 (YES in S12), an inquiry is made as to whether support data for correction is necessary (S12).
13). If the user inputs that the support data is unnecessary via the input unit 1 (NO in S13), the dictionary information is corrected by the dictionary information correcting unit 23 as follows (S15).

That is, the user looks at the contents of the dictionary data 30 shown in FIG. 17 displayed on the output unit 2 and obtains information on a headword that the user wants to correct via the input unit 1 (headword 31, translated word 3).
2A or 32B)
As shown in FIG. 17, only the information clicked by the dictionary information correcting unit 23 is extracted. For example, translation 32A
Is clicked "Please correct the translation."
Is displayed, and the user operates the input unit 1 to correct the contents of the edit box 95, and the corrected contents are reflected on the dictionary data 30 via the dictionary information correcting unit 23.

On the other hand, when it is instructed that the support data for the correction is required in the correction of the dictionary data 30 for each headword (YES in S13), the following processing (S14)
Is performed.

For example, in response to a user's request, the frequency information creation instructing unit 26 obtains the appearance frequency 4B at the time of character string extraction via the frequency information creation unit 290 of the auxiliary data creation unit 29,
By presenting as the frequency 93 shown in FIG. 17, it is possible to give a user a criterion for determining whether or not to register the corresponding headword 90 as the dictionary data 30.

When the translation device 501 uses the dictionary data 30, the dictionary lookup instructing unit 27 and the translation instructing unit 28 can be specified by the user to present desired support data.

For example, in FIG. 17, two dictionary reference buttons 96 and a translation button 97 are provided as reference information reference buttons.
One is shown. When any one of the displayed headwords 31 is clicked and selected and the dictionary lookup button 96 is pressed, the dictionary lookup instruction unit 27 instructs the dictionary lookup result creation unit 291 of the auxiliary data creation unit 29. The designated headword 31 is searched from existing dictionary data in the data storage unit 3, and the search result is displayed as support data.

Similarly, when translation button 97 is pressed, translation instructing section 28 causes auxiliary data creating section 2 to operate.
Nine translation result creation units 292 are instructed to operate, and the translation results of the specified headword 31 are displayed. These dictionary lookup results or translation results are displayed in the result window 98. FIG. 17 shows a result window 98 when the translation button 97 is pressed. Therefore, the user can easily determine whether or not to finally register the corresponding headword 31 in the dictionary data 30 by referring to the contents of the result window 98.

On the other hand, when the content correction is not instructed for each headword (NO in S12), headword recognition unit 5
An inquiry is made as to whether or not to correct the application of the various rules in (S16). In this case, unless a modification of the application of the rule is instructed, the series of processes is terminated. When the application of the rule is instructed to be corrected (YES in S16), the following processing is performed.

The application of each section of the headword recognition section 5 described above is referred to as application of a rule. These rules are based on the type of the input document data 40 and the created dictionary data 30.
There are cases where the application is appropriate and cases where it is inappropriate depending on the purpose of use. For example, if the numeric string in the document is the document data 40 describing a product used as a part of the model name instead of the version number, it is better not to apply the numeric string removing unit 8 described above. Therefore, if the user determines that the application is inappropriate in the application rule by the headword recognition unit 5 based on such criteria (S17)
Is YES, the application rule instructing unit 24 is activated (S1).
8).

The application rule designating section 24 displays a window on the output section 2 as shown in FIG. 18, and presents various rules in the headword recognition section 5. Here, for example, a list of application rules by the parentheses symbol removal / separation unit 6, the specific symbol division unit 7, and the word count check unit 10 is a rule list area 1.
12, an application check area 110 is provided for checking whether or not to apply in accordance with these various application rules. In this case, the rules displayed in the window are not limited to those shown in FIG.

In the application check area 110, for example, for each rule currently applied, the application button of the corresponding application check area 110 is presented in a checked state (●). If the user does not want to apply a rule,
By clicking the apply button in the corresponding application check area 110 and releasing the check state (○), it is possible to avoid applying this rule in the headword recognition unit 5.

On the other hand, if the presence or absence of the application of the rule displayed on the display window of FIG. 18 is not inappropriate (NO in S17), the character string 4A extracted from the document data 40 in accordance with the applied rule is inappropriate. If the user determines that an extracted character string may occur (YES in S19),
The suspension rule instructing unit 25 is applied (S20).

The user independently determines not only a rough instruction by the application rule instruction unit 24 as to whether or not to apply the above-mentioned rule but also each headword 31 which is a character string extracted using the rule. There are rules that need to be done. Therefore, the reservation rule instructing section 25 not only specifies whether or not to apply each rule of the headword recognizing section 5 but also finalizes each headword 31 which is a character string extracted by applying each rule. In the display window of FIG. 17, a registration button area 111 is provided for instructing that a registration button 94 be provided for each headword 31 so that the user can specify whether or not to register the dictionary data 30 in the dictionary data 30.

As shown in FIG. 18, the reservation rule designating section 2
5 is to add a registration button 94 to a rule having a high possibility of recognition error to urge the user to confirm the final registration in the dictionary data 30 for each headword 31.
The registration to which the rule is applied can be temporarily suspended.

For example, as shown in FIG. 17, a registration button 94 is provided for the headword 31, so that the user can instruct whether or not to register by a simple click operation of the input unit 1. At first, a registration button 94 is given in a check state (●), so if the user wants to cancel the final registration of the corresponding headword 31 in the dictionary data 30, the registration button 94 is given. Is clicked through the input unit 1 to release the check state (○), the corresponding headword 31 is not registered in the dictionary data 30.

[0194] The results of the above-described creation of each unit in the auxiliary data creation unit 29 are stored in the data storage unit 3 as various data 80, and are presented via the output unit 2 in response to a user request. It is also possible to obtain a final decision of the user and input an instruction to determine whether or not to register the entry word 31 in the dictionary data 30.

Further, by allowing the user to customize the support contents of the user support section 22, the burden on the user can be reduced, and an efficient dictionary creation system can be provided for each user.

A series of procedures for creating a dictionary according to the present embodiment is realized by a program according to the above-described flowcharts for causing the dictionary creation processing to function. This program is stored in a computer-readable recording medium.
Hereinafter, this is referred to as a program recording medium. In the present embodiment, the program recording medium may be a memory (not shown) required for the CPU 400 shown in FIG. 2 to perform the processing, such as a ROM itself. In addition, the medium access device 34 of FIG.
May be readable by inserting the program recording medium. In any case, the program recorded on the program recording medium is
It may be configured to be read and executed by being accessed by the CPU 400. Alternatively, in any case, the program may be read from the program recording medium, and the read program may be temporarily downloaded to a program storage area (not shown) in CPU 400 and then executed by CPU 400. .
It is assumed that the download program is stored in the main unit in FIG. 2 in advance.

Here, the above-mentioned program recording medium is
A recording medium that is configured to be separable from the main unit, such as a tape system such as a magnetic tape or a cassette tape, a magnetic disk system such as a flexible disk or a hard disk, or a CD.
-Disk system of optical disks such as ROM / MO / MD / DVD, card system such as IC card (including memory card) / optical card, mask ROM, EPROM, EEPROM
Any of a fixed system including a OM and a semiconductor memory such as a flash ROM for holding a program may be used.

In the present embodiment, since the apparatus shown in FIG. 2 can be connected to a communication network including the Internet using the input device 100, the program recording medium stores the program from the communication network in FIG.
It may be a medium that holds the program fluidly so as to be downloaded to the device. When the program is downloaded from the communication network as described above, the download program may be stored in the memory of the apparatus main body in FIG. 2 in advance, or may be installed from another recording medium. .

The contents stored in the above-mentioned program recording medium are not limited to programs, but may include data.

The embodiments disclosed this time are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

[Brief description of the drawings]

FIG. 1 is a diagram showing a functional configuration of a dictionary creation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a hardware configuration of a dictionary creation device according to the embodiment of the present invention.

FIG. 3 is a flowchart showing a schematic processing procedure for creating a dictionary using each function of FIG. 1;

FIG. 4 is a diagram showing an example of contents stored in a data storage unit of FIG. 2;

FIG. 5 is a diagram showing an example of the input document data of FIG. 4;

FIG. 6 is a processing flowchart of a character string extracting unit in FIG. 1;

FIG. 7 is a diagram showing extracted information obtained by the processing of FIG. 6;

FIG. 8 is a processing flowchart of a dictionary information creating unit of FIG. 1;

FIG. 9 is a diagram illustrating an example of dictionary data obtained by the processing of FIG. 8;

10 is a block diagram of a translation device that uses the dictionary data created by the dictionary creation device of FIG. 1 as a part of dictionary data referred to for translation.

FIG. 11 is a processing flowchart of a headword recognition unit of FIG. 1;

FIG. 12 is a processing flowchart of a headword recognition unit of FIG. 1;

FIGS. 13A and 13B are processing flowcharts of the headword recognition unit of FIG. 1;

FIG. 14 is a processing flowchart of a dictionary information correction unit in FIG. 1;

15 is a diagram illustrating an example of a processing result of the translated word full-width unit of FIG. 14;

FIG. 16 is a diagram illustrating an example of a processing result of the translated word information extraction unit in FIG. 14;

17 is a diagram showing an example of a display screen by the user support unit of FIG.

18 is a diagram showing another example of the display screen by the user support unit of FIG.

[Explanation of symbols]

1 input section, 2 output section, 3 data storage section, 4 character string extraction section, 5 headword recognition section, 17 dictionary information creation section, 18 dictionary information correction section, 22 user support section, 29
Auxiliary data creation unit, 30 dictionary data, 33 external storage medium, 34 medium access device, 40 input document data, 44 extraction information, 50 rule data, 500 dictionary creation device, 501 translation device.

Claims (33)

[Claims] An existing dictionary used for translating from a first language represented by a first type of character to a second language represented by a second type of character is not included in the existing dictionary. A dictionary creation device for creating dictionary information in which a registered headword is registered, comprising: an input unit for inputting information including a document in which a plurality of types of characters are mixed; and a document input from the input unit. A character string extracting unit that extracts one or more character strings composed of the first type of character and outputs extracted information in which the extracted character strings are registered; and a character string extracting unit that extracts the extracted information output from the character string extracting unit. A dictionary information creating unit that creates the dictionary information by using the extracted character string as the entry word and adding information about the entry word as entry word information corresponding to each of the entry words. 2. The one or more processing units for adjusting extraction of a character string so that the extracted character string is appropriate to be registered as the headword in the dictionary information. Having an adjustment processing unit having
The dictionary creation device according to claim 1. 3. The dictionary creation device according to claim 2, wherein the adjustment processing unit includes a parenthesis symbol processing unit that processes a character string before and after the parenthesis symbol in the extracted character string. 4. The adjustment processing unit includes a specific symbol division processing unit that divides the extracted character string before and after a specific symbol in the extracted character string to obtain the individual extracted character strings. 3. The dictionary creation device according to 3. 5. The dictionary creation device according to claim 2, wherein the adjustment processing unit includes a number string removal processing unit that removes a number string in the extracted character string. 6. The layout processing unit according to claim 1, wherein the adjustment processing unit includes a layout information processing unit that divides the extracted character string into individual extracted character strings based on information indicating a layout related to the document in the extracted character string. 6. The dictionary creation device according to any one of 2 to 5. 7. The adjustment processing unit, when the number of words in the extracted character string is equal to or less than a predetermined number,
7. The dictionary creation device according to claim 2, further comprising a word number processing unit that suppresses registration of the extracted character string in the extraction information. 8. The method according to claim 1, wherein the adjusting unit is configured to register the extracted character string in the extraction information when a length of the extracted character string including the predetermined number of words or less is equal to or less than a predetermined length. The dictionary creation device according to claim 7, further comprising a character string length processing unit for suppressing. 9. The method according to claim 1, wherein when the uppercase and lowercase characters are mixed in the extracted character string including the predetermined number of words or less, the adjustment processing unit keeps registration of the extracted character string in the extracted information. 9. The dictionary creation device according to claim 7, further comprising a character processing unit. 10. The adjustment processing unit determines a part of speech of the extracted character string based on a character string in the document subsequent to the extracted character string, and determines the extracted information according to whether the extracted character string corresponds to a predetermined part of speech. The dictionary creation device according to any one of claims 2 to 9, further comprising a part-of-speech processing unit that determines whether to delete the registration of the extracted character string in (1). 11. The dictionary creation device further comprises an auxiliary data creation unit that creates auxiliary data used to assist in creating the dictionary information, wherein the adjustment processing unit is configured to execute the auxiliary data creation unit 3. The adjustment processing is performed with reference to the created auxiliary data.
11. The dictionary creation device according to any one of claims 10 to 10. 12. The adjustment processing unit, wherein the auxiliary data creation unit includes a frequency information creation unit that counts the number of times the character string is extracted from the document for each of the extracted character strings and creates frequency information. The extracted character string in which the number of extractions is determined to be less than a predetermined number based on the frequency information created by the frequency information creating unit includes a frequency information processing unit that deletes the extracted information from the registration of the extracted information. The dictionary creation device according to claim 11. 13. The adjustment data creating unit, further comprising: a dictionary lookup result creating unit configured to search the extracted character string from a dictionary group including the dictionary information and the existing dictionary to create a search result; A dictionary lookup unit that suppresses registration of the extracted string in the extraction information when the search result created by the dictionary lookup result creation unit indicates that the extracted string is registered in the dictionary group. The dictionary creation device according to claim 11, further comprising a processing unit. 14. The translation data creation unit according to claim 11, wherein the auxiliary data creation unit includes a translation result creation unit that translates the extracted character string with reference to the dictionary group to create a translation result. Dictionary creation device. 15. The adjustment processing unit includes a first translation processing unit that maintains registration of the extracted character string in the extraction information when the translation result created by the translation result creation unit is not appropriate. , Claim 1
4. The dictionary creation device according to 4. 16. When the translation result created by the translation result creation unit indicates the extracted character string, the adjustment processing unit suppresses registration of the extracted character string in the extraction information. Comprising a processing unit,
The dictionary creation device according to claim 14. 17. The dictionary creation device according to claim 1, wherein the dictionary information creation unit includes a dictionary information correction unit for correcting the dictionary information. 18. The headword information includes a translation of the corresponding headword, and the dictionary information correction unit converts the character of the headword corresponding to the translation into a full-width character and registers it. The dictionary creation device according to claim 17, comprising: 19. The headword information includes a translation of a corresponding headword, and the dictionary information correction unit includes a translation word determination unit that determines and adds the translation of the headword from the document. Or the dictionary creation device according to 18. 20. The headword information includes a translation of the corresponding headword, and the dictionary information correction unit is configured to determine whether a character string is similar in at least one of the headwords in the dictionary information. 20. The dictionary creation apparatus according to claim 17, further comprising a translated word information unifying unit for assigning the same translated word to the one or more headwords. 21. An information correction unit for correcting the contents of the dictionary information created by the dictionary information creation unit as desired in accordance with information input from the outside via the input unit, The dictionary creation device according to claim 1. 22. The dictionary according to claim 21, wherein the information correction unit includes an external designation unit for designating the contents of the adjustment processing by the adjustment processing unit in accordance with an external instruction via the input unit. Creating device. 23. The external designation unit variably designates whether or not to apply the one or more processing units of the adjustment processing unit in accordance with an external instruction via the input unit.
The dictionary creation device according to claim 22. 24. The information correction unit, for each of the one or more processing units of the adjustment processing unit, finally converts the headword in the dictionary information to which the processing unit is applied, into the dictionary information 3. A registration determining unit for determining whether or not to register in the input unit via the input unit.
24. The dictionary creation device according to any one of 1 to 23. 25. The dictionary creating apparatus according to claim 22, wherein the creation data by the auxiliary data creation unit is presented at the time of designation by the external designation unit. 26. An existing dictionary used for translating from a first language represented by a first type of character to a second language represented by a second type of character, which is not used in the translation. A dictionary creation device for creating dictionary information in which a registered headword is registered, comprising: an input unit for inputting information including a document in which a plurality of types of characters are mixed; and a document input from the input unit. A character string extracting unit that extracts one or more character strings composed of the first type of character and outputs extracted information in which the extracted character strings are registered; and a character string extracting unit that extracts the extracted information output from the character string extracting unit. A dictionary information creating unit that creates the dictionary information by using the extracted character string as the entry word and assigning information about the entry word as entry word information corresponding to each of the entry words; Said letter The content of the information, and an information correction unit for correcting as desired in accordance with information inputted from the outside through the input section, a dictionary creation device. 27. The character string extraction unit includes an adjustment processing unit for adjusting extraction of a character string so that the extracted character string is appropriate to be registered as the headword in the dictionary information. The dictionary creation device according to claim 26. 28. The information processing apparatus according to claim 27, wherein the information correction unit includes an external designation unit for designating the content of the adjustment processing by the adjustment processing unit in accordance with an external instruction via the input unit. Dictionary creation device. 29. The dictionary creation device further includes an auxiliary data creation unit that creates auxiliary data used to assist in creating the dictionary information, wherein the adjustment processing unit is configured to execute the auxiliary data creation unit 3. The adjustment processing is performed with reference to the created auxiliary data.
The dictionary creation device according to 7 or 28. 30. The dictionary creating apparatus according to claim 29, wherein at the time of designation by said external designation section, data created by said auxiliary data creation section is presented. 31. A translation device for performing translation processing from a language represented by a first type of character to a language represented by a second type of character, wherein the translation device is extracted from an existing dictionary and a document in which a plurality of types of characters are mixed. One or more headwords that are not registered in the existing dictionary, and dictionary information having information on the headword as headword information for each of the headwords is stored. A translation device, comprising: a dictionary storage unit configured to refer to information in the dictionary storage unit for the translation process. 32. A dictionary creation method for creating dictionary information used for translating a first language represented by a first type of character into a second language represented by a second type of character, comprising: An input step for inputting information including a document in which characters are mixed, and the first step based on the document input from the input step
A character string extracting step of extracting one or more character strings composed of different types of characters and outputting extracted information in which the extracted character strings are registered; and the extracted characters of the extracted information output from the character string extracting step A dictionary information creating step of creating dictionary information by assigning information relating to the headword as headword information corresponding to each of the headwords in a column. 33. A computer-readable recording medium storing a dictionary creation program for causing a computer to execute a dictionary creation method for creating dictionary information used for translating from a first language to a second language. The dictionary creation method is a dictionary creation method for creating dictionary information used for translating a first language represented by a first type of character into a second language represented by a second type of character. An input step for inputting information including a document in which different types of characters are mixed; and the first step based on the document input from the input step.
A character string extracting step of extracting one or more character strings composed of different types of characters and outputting extracted information in which the extracted character strings are registered; and the extracted characters of the extracted information output from the character string extracting step A dictionary creating program for creating dictionary information by providing a column as a headword and providing dictionary information by providing information on the headword as headword information corresponding to each of the headwords. Possible recording medium.