JPS6244875A - Machine translator - Google Patents

Abstract

PURPOSE:To efficiently execute a translation editing work by retrieving proposed translated words regarding the basic format of an original word or an original word obtaining the original form of a translated word as a heading by referring a translation dictionary part and displaying the retrieved word to correct the translated word. CONSTITUTION:An original inputted from an input part 1 is stored in an original storing part 2. A translation part 4 successively executes the machine translation processing of the original on the basis of knowledge information of the translation dictionary part 5. An edition control part 3 displays the original stored in the original storing part 2 and the translation stored in a translation storing part 6 correspondingly each other. If an unsuitable translated word is included in the translation at the display of the original and its translation on a display part 8 after the end of the translation processing, proposed translated words regarding the basis form of the original word or the original word obtaining the original form of the translated word as the heading are retrieved and displayed to correct the translated word.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a machine translation device that can improve the efficiency of translation and editing work.

[Technical background of the invention and its problems]

2. Description of the Related Art Recently, machine translation devices that automatically machine translate an input original text using a computer and obtain the translated text have been attracting attention. For example, input a Japanese sentence and get its English translation,
Various attempts have also been made to develop machine translation devices for ancient natural songs that input an English sentence and obtain its Japanese translation.

This type of device basically performs morphological analysis, syntax analysis, etc. on an input original text, and divides it into predetermined processing units, such as imperial edicts (phrases). Then, the translation dictionary is searched for each processing unit to find translation words (translation phrases) corresponding to each processing unit, and these translation words (translation 3R phrases) are combined according to predetermined translation rules to find the translation. Ru.

However, since semantic interpretation technology for natural languages has not been sufficiently established, it is extremely difficult to reliably obtain an appropriate translation using the above-mentioned machine translation.

Specifically, for example, the statement (find) is (understand,
One original word often has multiple FJ4 edict candidates, as it has translations such as feel, find, find out, solve, destroy, expose, discover).

Therefore, depending on which translation is selected, the linguistic expression of the translated text changes greatly, and the semantic nuance of the translated text is likely to be misaligned.

Conventionally, a plurality of translation word candidates for each processing unit (original word) are presented to the operator, and a translated sentence is constructed using the translation word selected and designated by the operator as the most appropriate.

However, even when a translated text is obtained in this way, the translated text often still contains inappropriate translated expressions. Therefore, as a post-editing process for the translated text, there remains a need for the operator to correct the inappropriate translated word into an appropriate translated word.

[Purpose of the invention]

The present invention has been made in consideration of these circumstances, and its purpose is to efficiently perform translation editing work on translated texts by effectively utilizing the knowledge information stored in the translation dictionary section. The purpose is to provide a machine translation device.

[Summary of the invention]

The present invention provides a machine translation device for translating an original text using knowledge information stored in a translation dictionary section, displaying the resulting translated text and the original text, and performing translation editing on the translated text. Indicate the translation 8B to be corrected in the translated text or the original word in the original text from which the translated word was obtained, and use the translation dictionary section to select translation word candidates whose headword is the basic form of the original word or the original word from which the original form of the translated word is obtained. You can refer to it, search it, display it, and use it to correct the translation.

At the same time, we grammatically analyze the translated sentences that include the target word to be corrected, and select the appropriate translation from among the above candidate translation words! The candidate is transformed and generated according to the above grammar analysis result so that it becomes an appropriate linguistic expression in the translated sentence, and the translated word obtained by this transformed generation is replaced with the translated word to be corrected in the translated sentence to correct it. This is how it was done.

〔Effect of the invention〕

Thus, according to the present invention, when the translation process is completed and the translated text and the original text are displayed, if the translated text contains an inappropriate translation 8h, it is used to validate the knowledge information stored in the translation dictionary. It is possible to use it for correction. In addition, the translation word candidates selected according to the results of grammatical analysis of the translated text are transformed and generated so as to be expressed in the original language in the translated text, and the translated word is corrected, so that the processing can be performed efficiently.

Therefore, it is possible to simplify the translation and editing work, and to obtain a translated text with an appropriate linguistic expression easily and efficiently, which brings about great practical effects.

[Embodiment of the Invention] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

In this embodiment, an English text is input and translated into Japanese by machine. FIG. 1 is a schematic diagram of the apparatus of this embodiment.

In FIG. 1, reference numeral 1 denotes an input section 1 consisting of a keyboard and the like. The English sentence inputted from the input unit 1 or the like is stored in the original text storage unit 2 as an original text to be subjected to translation processing.

The translation unit 4 uses the knowledge information for translation processing stored in the translation dictionary unit 5 in advance under the control of the editing control unit 3 to sequentially convert the original text stored in the original text storage unit 2 into a predetermined 51 !Machine translation processing is performed at the 1st position.

The knowledge information stored in the translation dictionary section 5 includes, for example, a rule/irregular change dictionary 5a, a word (translation) dictionary 5h, a qI dictionary 15G for unconnectable part-of-speech strings, a translation dictionary 5d, etc. . Using such knowledge information, a Japanese sentence (YU), which is a translated sentence obtained by machine translating the original sentence, is managed to be associated with the original sentence and sequentially stored in the translated sentence storage unit 6.

The editing control section 3 drives the display control section 1 to display the English sentence stored in the original text storage section 2 and the translated text storage section 6.
The Japanese sentences stored in the Japanese sentences are correlated with each other and displayed simultaneously on the display section 8, and the translated sentences are subjected to post-editing processing. This post-editing process is executed according to control information input from the input section 1, for example, by referring to knowledge information stored in the translation dictionary section 5, as will be described later.

The original text completed after such post-editing processing (
The translation text (Japanese iiM l! ()) for the English text is output as a hard copy by the printing unit 9.

By the way, FIG. 2 shows an example of the structure of the keyboard that constitutes the input section 1. As shown in FIG. This keyboard includes, for example, in addition to key details 1a for inputting character data, keys 1b for Ill translation instructions, a group of editing keys 1c, a group of function control keys 1d,
It is comprised of a group of keys 18 for controlling the cursor in the display section 8, and the like.

Further, FIG. 3 shows an example of the configuration of a display screen for the original text and translated text on the display section 8. As shown in FIG. Here, the display screen is divided into three parts: a translation editing area 8a on the L part of the screen, an original text display area 8b1 on the left side of the screen, and a translated text display area 8C on the right side of the screen. In this original text display area 8b, the original text storage section 2
The input original texts stored in are sequentially displayed, and the translated text stored in the translated text storage section 6 is displayed in the translated text display area 8C in correspondence with the original text from which the translated text was obtained. Furthermore, Translation Editorial Area 1
i1!8a displays information necessary for the translation process, such as translation word candidates searched from the translation dictionary section 5 and provided for the translation process.

FIG. 4 shows the basic operation sequence of the embodiment apparatus configured as described above. According to such an operation sequence, the editing control section 3 determines the translation end information given from the translation section 4 and various key information input from the input section 1, and interactively controls the translation/editing process. .

That is, the editing control unit 3 monitors the translation processing status in the translation unit 4, and in step A), when it detects that the translation processing of one original text in the translation unit 4 is completed, the editing control unit 3 changes the translation obtained by the translation processing to the translated text. The translated text is stored in the storage unit 6 and displayed on the display unit 8 in correspondence with the original text (step B).

Further, when the translation process is continuing in the translation unit 4 or when the translation process is not being performed, the editing control unit 3 determines the key information input from the input unit 1 (steps C, D, and E). ,F). If the input key information is a "translation instruction key" (step C), the editing control section 3 gives the input original text stored in the original text storage section 2 to the translation section 4, and starts the translation process (step C). G).

Further, if the input key information is a "character key" (step D), store the character code indicated by the character key in the input buffer (step 1), and store the character code in the original text register tlil'M2. , the character pattern is displayed on the display section 8 (step I). The original text input from the input section 1 is displayed by displaying each character pattern of the character code stored in this manual buffer.

Furthermore, if the input key information is the rm collection key j (step E), the f/a collection process corresponding to the edit key is executed on the translation (step J). Similarly, if the input key information is "Function key J" (step F
), the process corresponding to that function key is executed (step K).

If the key information is not input, or if the input key information is other than the above-mentioned "key", other processing, such as outputting a hard copy of the translation obtained in the translation storage section 6, is performed. It will be done.

With such an operation sequence of the editing control unit 3, for example, the operator can press the character input key group 1a on the keyboard.
When characters are input by operating , the character information is sequentially transferred to the input buffer and sequentially stored in the original text storage section 2 as an original text to be subjected to translation processing (steps D and H).

The input original text is the original text display area 8b of the display section 8.
(Step I).

When the translation instruction key 1b is operated at any point in character input, for example, at the end of inputting one sentence, translation processing for the input original text stored in the input horn buffer is started according to the key human power information (step C ,G). When the translation process is completed, the resulting translation is displayed in the translation display area 8C of the display section 8 (steps A and B).

If it is necessary to edit the input original text, for example, operate the cursor control key group 1e to move the cursor to the corrected part and make corrections while inputting the original text by operating the character input keys 1Y1a.・Edit keys for inserting, deleting, etc. !
By manipulating ¥10, the editing process is executed (steps E and J).

In this way, the basic operations of machine translation processing are controlled.

By the way, the translation process of the original text is executed as follows, for example.

FIG. 5 shows an example of the processing sequence.

The translation unit 4 first analyzes the word 8B form of the original text to be translated using the regular/irregular inflection word 15a (step P). Through this morphological analysis, for example, an original word that has undergone a conjugation change or an inflection change is converted into its original form (basic form).

Next, for each original word of the original text that has been morphologically analyzed as described above,
Information such as part-of-speech information and translated words is obtained using the translation fJ n 15 b (step Q). This process is performed by searching the translated word still image 5b using the original 8B as a headword.

Thereafter, the connectability of the translation word candidates is verified according to the dictionary searched information (step R).

This verification is performed with reference to the unconnectable part-of-speech string rule 5C, and is repeated until a consistent parsing result (arrangement of part-of-speech strings of translation 8n candidates) is obtained. Through this syntactic analysis, we can understand the arrangement structure of the parts of speech of the original language that makes up the original sentence,
The relationship, tense, etc. will be required for this dependency.

Thereafter, the translation dependency converts the parsed structure of the original sentence into the syntactic structure of the translated sentence using the dictionary 15d, and generates a translated sentence consisting of a list of translation word candidates for each source language (step S). At this time, each H
g candidates are subjected to conjugation transformation and 88-tail transformation processing, and the translated text is made into an appropriate linguistic expression.

The translated text for the original text obtained through such translation processing is stored in the translated text storage section 6 and displayed in correspondence with the original text. Incidentally, when a plurality of types of translations for one original text are obtained, for example, only one translation may be displayed, and at the same time, the existence of other translations may be identified and displayed.

By the way, if an instruction is input using the edit key after the translated text obtained in this way and the original text from which the translated text has been obtained are displayed on the display section 8, the translation l! of the translated text will be changed! East processing is performed as follows.

The translation editing process for this translated text is carried out at 8B (original word, original phrase,
This is achieved by performing processing corresponding to the operated edit key on the translated words, station words and phrases). Specifically, you can: ■ Insert a character before the cursor position by operating the insert key; ■ Delete the character string in the range indicated by the cursor by operating the delete key; ■ Insert a character before the cursor position by operating the move key. ■ Move the character string in the range specified by the cursor, ■ Disable each editing function specified by each of the above keys by operating the Cancel key, ■ Modify the word or phrase indicated by the cursor by operating the keys. The modification consists of translation editing processing such as displaying candidates.

Further, when the above-mentioned function control keys are operated, the following functions are provided, which are used for the translation of the above-mentioned translated text.

That is, ■ When the translation display key is operated, the translated word will be displayed for the word in the translated text that the cursor points to. ■ When the dictionary display key is operated, the translated word will be displayed for the word in the original text that the cursor is pointing to. Displays the contents of the translation dictionary with the word as heading 8R; ■ When the dictionary registration key is operated, the character string indicated by the cursor is registered in the dictionary as a new word/idiom; ■ When the dictionary deletion key is operated, the dictionary Deregistering a registered new word/idiom. ■ When the partial translation key is operated, this is indicated by the cursor and is realized by a function such as displaying a partial translation for the original text for which translation processing has failed.

In addition, the instructions for character strings (edicts) etc. using the cursor mentioned above are as follows:
This is performed by moving the cursor on the display screen by operating the cursor movement keys and varying the cursor size by using the cursor control keys.

By using these various functions, post-processing such as correcting the translation of the above-mentioned translation is performed interactively.

The above are the basic translation processing functions that this device has. Here, the feature of this device is that when correcting an inappropriate translated word in the translated sentence, the v
Riii! The point is that knowledge information such as ivI book 5b can be actively used.

That is, in the state where the original text and the translated text are displayed, for example, as shown in FIG. At this time, translation candidates for the eight originals and translation 3B candidates for the original word from which the translation was obtained are obtained from the translation dictionary section 5, and these are displayed as shown in FIG. 6(b). Then, when one translation candidate is selected from the displayed translation candidates, the selected translation is transformed and generated so that it becomes an appropriate linguistic expression in the translated sentence, and is shown in FIG. 6 (C). As shown, the present invention is characterized in that the translated text is corrected by replacing the translated word to be corrected in the original text.

FIG. 7 shows the processing sequence, and the flow will be explained with reference to FIG. 6.

Now, when the translated word correction mode is set and a word in the original or translated text is specified with the cursor on the display screen, the first thing to do is to check whether the specified middle part is the original word in the original text or the [i] in the translated text. It is determined whether there is. Single 8R indicated by the cursor
is the original language displayed in the original language display area 8b,
First, the original language is identified (step a).

On the other hand, when a translated word is indicated by the cursor, the translated word is first specified (step b).

Then, the original word from which the translated word was obtained is identified from the original text corresponding to the translated text that includes the specified translated word (step C).
. The original word corresponding to the translated word is specified by, for example, referring to a table showing the word correspondence between the original text and the translated text, as shown in FIG.

Once the translated word to be corrected and the original word from which the translated word was obtained are identified in this way, the original i! The H form is grammatically parsed and the parsing results are stored (step d). Then, the original form conversion process of the specified original word is performed (step e). This original form conversion process is performed by selecting the original form if there are multiple possibilities.

For example, I found the host.
l, J in the original sentence. When the original word J is specified, this original word has the possibility of being the lq existential form of "found: iQ stand (verb)" and the past of "find; find (verb)". It could be a shape. Therefore, first, one of them is specified, for example, according to the result of semantic analysis of the original text.

If the specified original g is specified as the past tense of [1ndj, then the specified original form is converted into the specified form by the original form conversion process (including irregular conversion rules) from past tense to present tense two basic forms. Original form fJ f four+rl J (F) Find the original form ``findj.For example, in the original form conversion process, [boy's j in the form of <possessive form> → <base form> is converted to ``boy J''. Processing such as converting 1rooino j to "00" in the form of <progressive form> → <basic form>, and further processing such as converting 1rooino j to "00" in the form of <progressive form> → <basic form>, and furthermore, processing of "better, besB" in the form of <comparative form/superlative form> → <basic form> “good”
This includes processing to convert to j.

After that, a dictionary search is performed using the original form of 8 as a heading candy for the specified original (step f), and the translation BH candidates found by this dictionary search are, for example, as shown in FIG. 6(b). displayed (step 0).

If an appropriate translation expression exists among the translation candidates displayed in this manner, the translation candidate is designated (step h). Then, transformation processing is applied to this designated translation 8B candidate according to the grammatical analysis results of the translated text, and the translated text is transformed so that it becomes an appropriate original expression (
Step 1).

For example, the past tense is ``When the translation candidate ``discover'' is selected from among the translation candidates that punish Haraichi called f'1ndJ, the translation to be corrected in the previously sought translation was found r.'' Therefore, the translated word ``discovered'' is generated. This modified generation of translated words may be performed, for example, according to the parsing grammar rules as illustrated in FIGS. 9(a) to 9(e).

Specifically, the translation 8n “r found” is “found/
It is grammatically analyzed as Mashi・TaJ and its structure is found (1)
Since it is analyzed as (stanza) + (style - desmas) + (tense - past), the translation candidate for "discovered" can be transformed and generated as "discovered" according to this grammar structure.

By replacing the translated word "discovered" obtained in this manner with the translated word "found" to be corrected in the translation (step J), the translation correction is completed.

In this way, according to the present device, information on translation candidates related to the translation to be corrected can be effectively refined by actively using the knowledge information stored in the translation dictionary.

That is, when executing the translation process, it is possible to select the translation candidates required from the translation dictionary each time and select the appropriate translation, but when the translation process is finished and the translated text and its translated text are obtained. At the time when the original text is displayed, the connection between the original text and the translated text and the translation dictionary is temporarily disconnected. Even in such a state, by specifying a word in the displayed original text or translated text and operating the dictionary display key, the translation dictionary 5 displays the translation word candidates with the original form of the original BB as the heading edict.
Since you can search and display from the translation dictionary, you can effectively use the knowledge information stored in the translation dictionary to improve the translation! This makes it possible to efficiently conduct operations in the east.

Moreover, the translation candidates selected according to the results of grammatical analysis of the translated text are transformed into appropriate linguistic expression forms in the translated text and then used for translation correction, so that the semantic information of the original text is well reflected and the words are expressed appropriately. Translations can be obtained easily and effectively.

Therefore, great practical effects can be achieved, such as simplifying translation and editing processing and making it more efficient.

It should be noted that the present invention is not limited to the embodiments described above. For example, IJ? j! ! The method of generating candidate transformations and the grammar analysis method for generating transformations can be set according to the specifications. Further, here, translation and editing are performed using cities as processing units, but it is also possible to use saishun as the processing units. Also, all translations for the specified word BB
Of course, it is also possible to display the R candidate from the translation dictionary and use it for correcting the translated word. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of drawings]

The figures show a device according to an embodiment of the present invention; FIG. 1 is a schematic configuration diagram of the device, FIG. 2 is a diagram showing an example of the configuration of a keyboard,
Fig. 3 shows an example of a display screen, Fig. 4 shows a basic operation sequence, Fig. 5 shows a translation processing sequence, Fig. 6 shows an example of a dictionary display, Fig. 7 8 is a diagram showing a processing sequence for dictionary display, FIG. 8 is a diagram showing an example of the structure of a table showing word correspondences, and FIG. 9 is a diagram showing a grammar analysis method. 1... Input section, 2... Original text storage section, 3... Edit control section, 4... Translation section, 5... Translation dictionary section, 6...
Translation storage unit, 7... Display control unit, 8... Display unit, 9
...Printing Department. Applicant's agent Patent attorney Takehiko Suzue---C Chu Mourning Group 1 button value）-5a----------------------
(Eat 5b t!;'it ii 1 梠---(,-)-5c 晔 i 6 y y san, l□1 1 ------- <1 shakusha irihiki- y)-5a5Figure 6Figure 7Figure 8

Claims (2)

[Claims] (1) In a machine translation device that displays a translated text obtained by translating an original text using knowledge information stored in a translation dictionary section and the original text, and performs translation editing on the translated text, a translated word to be corrected in the translated text. means for selectively instructing the original word in the original text obtained or a translated word to be corrected in the translated text; means for searching and displaying from the translation dictionary section, and when a translation word is selectively specified, searching from the translation dictionary section for translation candidates having the original word as a headword from which the original form of the translation word is to be obtained; A machine translation device characterized by: (2) In a machine translation device that displays a translated text obtained by translating an original text using the knowledge information stored in the translation dictionary section and the original text, and performs translation editing on the translated text, a translated word to be corrected in the translated text. a means for selectively instructing the original word in the original text obtained by the above-mentioned translation, or a translated word to be corrected in the translated text; a means for grammatically analyzing a translated text containing the translated word to be corrected; When instructed to do so, the translation dictionary section is searched and displayed for translation candidates that use the original form of the original word as the entry word, and when a translation is selectively specified, the original word from which the original form of the translated word is obtained is used as the entry word. means for searching and displaying translation word candidates from the translation dictionary section; means for selectively specifying the translation word candidates searched and displayed from the translation dictionary section; and means for selectively specifying the translation word candidates searched and displayed from the translation dictionary section; Means for generating a transformation so that an appropriate linguistic expression is obtained in the translation according to the result of grammatical analysis of the translation, and means for replacing the translation obtained by generating the transformation with the translation to be corrected. A machine translation device characterized by: