JPS58151677A - Editing method of translated sentence - Google Patents

Abstract

PURPOSE:To improve the operation efficiency, by correcting a sentence constructing element to e corrected only in one position to correct it in other positions automatically when it appears plural times in a text and making it unnecessary to input characters other than characters to be changed in the correcting operation. CONSTITUTION:A translated character editing device is provided with a CPU1, and the whole of the device is controlled by this CPU1, and text in the first language, a text translated into the second language, correspondence relations of sentence constructing element units between both texts, etc. are stored in a storage device 2. The text is displayed on a CRT3 connected to the CPU1, and the text correcting indication is inputted from an input device 4. When the sentence constructing element to be corrected appears plural times in the text of the storage device 2, it is corrected only in one position, and this element is corrected in other positions automatically under the control of the CPU1; and thus, it is unnecessary to input characters other than characters to be changed in the correcting operation, thereby shortening the operation time and improving the operation effciency.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for proofreading translated text in a machine translation system, and particularly to a method for correcting words and phrases in translated text.

When a machine translation system performs translation, the translation algorithm operates in the same way for the same butter/ in the original text, so the same mistranslations will occur over and over again. Therefore, it is important to have the ability to perform the same correction at multiple locations throughout the text. Conventional text editors are provided with a function of directly converting a specific character string ak specified by the user into another character string a' specified by the user. Normally, the number of characters input to replace the character string a with the character string a' is larger than the number of characters actually modified. In other words, even if you only need to correct some letters in a particular word or phrase, the letter may be included as part of a different word or phrase, so You must enter the entire word or phrase, or at least enough characters to distinguish it from different words or phrases. In extreme cases,
The length of the number of characters required manually to correct just one character is 1
It may be 0. Therefore, conventional text editors have the problem that the number of input characters increases more than necessary, reducing the editing efficiency.

This becomes a particularly serious problem when dealing with Japanese, which requires a lot of human effort per character.

An object of the present invention is to solve all of the above-mentioned problems and to provide an efficient translated text editing method [6].

In a machine translation system, the present invention requires that the syntactic elements (words, collocations, phrases, etc.) in a text be recognized by the system. The idea was to eliminate the need to input data each time. In other words, in the present invention, a certain syntactic element is included in multiple places in the text, and
If the user wants to modify all of them in the same way, the user can select one location and modify it by changing, pressing, or deleting characters, and then press fc.
Just press a specific function key. Since the system can identify the syntactic element that has been modified, it checks whether the same syntactic element as this one exists elsewhere and, if it does, applies the same modification to that element as well. Run automatically.

An embodiment of the present invention will be described below. As shown in FIG. 1, the editing device used in the method of the present invention includes a central processing unit 1 that controls the entire device and performs processing such as text correction, text in a first language and translation thereof in a second language. A storage device 2 for storing the text and the correspondence of syntactic elements between the two texts. CR, T showing the entire text
3. Human-powered device for inputting text correction instructions [consisting of 4
Ru.

This editing device is connected to a central computer that executes the entire translation program via a channel or communication line, and the text in the first language, the text in the second language, and the correspondence of syntactic elements between the two texts are processed by the central computer. I'll send it to you from here. Further, these takes are corrected using the functions of the editing device and then sent to the central computer.

The storage device 2 stores text in a second language as shown in FIG. A first text area 21 for storing text; and a second text area 22 for storing text based on the first word. Words in the text? Word table 231. A collocation table 232 that shows collocations in the text, a phrase table 233 that shows phrases in the text, and a modified word area 241 that stores all character strings of words before and after modification. Corrected collocation area 242 that stores character strings of collocations before and after modification. It consists of a modified phrase area 243 that stores the character strings of phrases before and after modification. word table 231
In the (collocation table 2322 phrase table 233), each row corresponds to one word (one collocation phrase) with 7'n-f:', and the starting address 2311 ( 2321, 2331) and final address 2
312 (2322, 2332), starting address 231 of the word (9 collocation phrases) in the second text area 22
3 (2323, 2333) and the final address 2314 (23
24, 2334) are stored.

Corrected word area 241 (corrected collocation area 242,
The modified phrase area 243)K stores information regarding the word (9 collocations) that was modified most recently in terms of time, and stores the character string 2411 (2411) that represents the word (9 collocations) in the first language.
421, 2431), a character string 2412 (2422, 243) representing the word (9 collocations) before correction in the second language.
2), and a character string 2413 (2423, 2433) representing the corrected word (nine collocations) in the second language.

FIG. 3 shows a board diagram of the input device 4. In the embodiment, the second language is Japanese, and characters such as kana and kanji are input using a tablet. As shown in the figure, the input device 4 includes a tablet 41 for inputting characters, a pen 42 . CfLT3
Cursor keys 43 for moving the upper cursor. Character insertion key 44 for inserting all characters into text. Character deletion key 45 for deleting all characters from text. Word modification key 46 for modifying the entire translated text of Ruru words. A translation of a certain collocation? A combination correction key 47 is provided for correcting the entire text, and a phrase correction key 48 is provided for correcting the translation of a certain phrase throughout the text. Note that keys not directly related to the functions of the present invention are omitted.

Next, an explanation will be given of an input operation when correcting the translation of a phrase to be used throughout the text. The user first selects one of the occurrences of the phrase, and modifies it using the tablet 41, pen 421, character insertion key 44, character deletion key 45, and cursor key 43. To change characters, move the cursor to the position of the character you want to change, and press Tablet 4.
This is done by specifying all desired characters on 1. Insertion of characters is performed by moving the cursor to the desired insertion position, pressing the character insertion key 44, and then specifying all desired characters on the tablet 41. A character is deleted by moving the cursor to the position of the character to be deleted and pressing the character deletion key 45. The processing flow of the central processing unit 1 for these 1-positive operations is shown in FIG. 4(a). In this way, when the user presses the 48 phrase correction keys after all corrections have been made to one of the phrases in question, the process shown in FIG. Corrections will be made automatically.

Referring to FIG. 4(a), the processing of the central processing unit 1 in response to operations for changing, inserting, and deleting characters will be explained. The central processing unit 1 stores the cursor position and the word double 231 . From the collocation table 232 or the phrase table 233, it can be determined which word, collocation, or phrase is to be modified (character change, extension/deletion, etc., collectively referred to as modification). Therefore, when a correction instruction is input, it is difficult to judge whether the correction is in the same word as the previous correction or not.
), if the correction is to a word different from the previous one, the character string in the first language of the word, the character interest rate before correction in the second language, and the first item 2411 of the corrected word area 241.
and set in the second item 2412 (102).

Similarly, if it is a correction to a collocation that is different from the previous one (103
), the character string in the first language of the collocation and the character interest rate before correction in the second language are set in the first item 2421 and second item 2422 of the four-word correction area 242 (104). Furthermore, if the modification is to a phrase different from the previous one (105), the character string in the first language of the phrase and the character string before modification in the second language are stored in the modified phrase area 243.
The first item 24312 and the second item 2432 are set (106).

Next, update the test text in the second language according to the correction instructions (107), and update the character strings in the second language of the words (9 collocations) of 1. 242. Third item 241 of corrected phrase area 243)
Set to 3 (2423, 2433) (108 (1
09,110)). Through the above processing, the text is updated and the character strings before and after the modification of the last modified word (nine collocations) are always retained.

Next, according to FIG. 4(b), the phrase correction key 48 is pressed.
The processing of the central processing unit 1 when ``is performed will be explained.

When the phrase modification key 48 is pressed, the first text area 21 is searched for the same character string as the first item 2431 of the modified phrase area 243 (111).

If the character string is not found (112), the process ends; however, if the character string is found, the entire next process is performed. First, referring to the phrase table 233, it is checked whether the searched character string is a phrase (113). If it is a phrase, the phrase table 233 and second text area 22 are further referred to, and all character strings in the second language of the phrase are extracted, and whether or not this matches the second item 2432 of the corrected phrase area 243 is determined. Check (114). If they match, the character string in the second text area 22 of the phrase is replaced with the third item 2433 in the modified phrase area 244 (115).

(9) By repeating the above process, the correction phrase area 243
All phrases that are the same as the phrase inside are corrected.

FIG. 4(b) shows the processing flow when the phrase correction key 48 is pressed, but when the word correction key 46 or the collocation correction key 47 is pressed, one phrase 1 in the figure 1 word 1
Alternatively, the processing equivalent to replacing 1 slow word with 1 is carried out in a big way.

According to the present invention, when a syntax element to be modified appears multiple times in a text, it is possible to modify only one location and automatically make the same modification to other locations. Also, in the first normal operation, there is no need to input characters other than the characters to be actually changed, and the operation efficiency is improved compared to the character string direct conversion function in the conventional text eater. Since the number of input characters can be kept to the minimum necessary, this is a great effect when dealing with all Japanese languages, which requires time to manually write characters.

[Brief explanation of the drawing]

Figure 1 shows the configuration of the editing device used in the method of the present invention. 2 is a diagram showing the constituent sounds (10) of the area of the storage device, FIG. 3 is a board diagram of the input device, and FIG. 4 is a processing flow diagram of the central processing unit in the present invention. 1...Central processing unit, 2...Storage device, 3...C
RT. 4... Input device agent Patent attorney Toshiyuki Usuda (11) No. 1 Hard No. Z Diagram

Claims (1)

[Claims] A step of storing a text in the first language, its translated text in the second language, and the correspondence between the syntactic elements of both texts, and storing the character strings before and after modification of the syntactic elements that were modified when the translated text is modified. and searching for the same syntactic element as the modified syntactic element throughout the text, and then performing the same modification for n? Features a translation editing method.