JPH0310372A - Display and processing system for original word/ translated word correspondence to technical term - Google Patents

Abstract

PURPOSE:To efficiently advance a tranalation work by displaying the corresponding information of the original word and its translated word of a technical term extracted by a technical term extraction processing part, when the translation is unsuccessful. CONSTITUTION:When a translation is unsuccessful due to an analytic error by an original syntax analyzing part 14 and a translated sentence generation error by a translated sentence generating part 15, an original word/translated word correspondence display processing part 17 edits corresponding information of the original word and its translated word with regard to a technical term extracted by a technical term extraction processing part 13, related to its original, and displays its corresponding information in an original word/translated word correspondence display area 23 in a screen of a display device 20. That is, the translated word of the original word related to the original being unsuccessful in the translation is displayed in the original word/translated word correspondence display area 23 only with regard to a technical term registered in advance in a translating dictionary 19. In such a way, at the time of editing after the translation, a translated sentence can be generated and inputted quickly and easily by referring to its original word/translated word corresponding information, and the work efficiency is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] This invention relates to a processing method for displaying technical terms corresponding to their original language and translation, which facilitates post-editing work in the event of translation failure in a machine translation system that performs translation using a computer.

In the event that translation of the original text fails, the purpose is to display the translated words for specialized terms in the original text and to provide a means to support post-translation editing work. A translation dictionary that stores dictionary information with information added thereto.

A technical term extraction processing unit that identifies whether or not each original word in the original text to be translated is a technical term based on the search results of the translation dictionary and extracts two technical terms, and when translation fails, The source language/translation correspondence display processing section that performs processing to display correspondence information between the original language of the technical term extracted by the technical term extraction processing section 2 and its translation is configured so as to be able to display (liff).

[Industrial Application Field] The present invention relates to a processing method for displaying technical terms corresponding to original and translated words, which facilitates post-editing work in the event of a translation failure in a machine translation system that performs translation using a computer.

[Conventional technology]

Although machine translation systems are being put into practical use, machine translation systems cannot accurately translate all documents. Therefore, post-editing work by a human translator is necessary, and a considerable amount of time is spent on post-editing work.

In particular, 2 For documents that contain many terms related to specialized fields or terms with special meanings that are not used in normal daily conversation, 9 If machine translation fails, a dictionary for those specialized terms may be required. Since two documents are required to be researched, a lot of time is spent on post-editing work performed by two people.

With conventional machine translation systems, there was no effective means to support post-editing work even if translation failed.

[Problem to be solved by the invention] In conventional machine translation systems, when translation fails, 2
Translators had to research appropriate translations of technical terms, etc. in documents using dictionaries, literature, etc. before translating them. Therefore, there is a problem in that post-editing requires a lot of effort and time.

In order to reduce this workload, it may be possible to automatically display the translated words for all the original words in the original text when translation fails.1 Normally, the original text where translation fails is often long. (Also, since the structure of sentences is often complex, if all translated words are displayed enumerated, one display becomes complicated and difficult to read.

The present invention aims to solve the above-mentioned problems, and when the translation of the original text fails, it displays translated words for terms in specialized fields specified in the dictionary in advance among the original words that appear in the original text, and allows post-translation editing work to be performed. The aim is to provide a means of support.

[Means to solve the problem] FIG. 1 shows an example of the configuration of the present invention.

In FIG. 1, 10 is a processing unit consisting of a CPU and memory, etc., 11 is a source human resource unit that inputs the original text to be translated, 12 is a source text morphological analysis unit that performs morphological analysis of the original text, and 13 is a unit that analyzes technical terms from the original text. 14 is a source text syntax analysis unit that analyzes the syntax of the original text;
5 is a translation generation unit that generates a translation based on the syntax analysis results;
16 is a translation output unit that outputs a translated text; 17 is a source language/translation correspondence display processing unit that displays correspondence information between the original language and the target word when translation fails; 18 is an input device such as a keyboard or an original text file; 19 is a source language or 19A is an original language dictionary, 19B is a translated dictionary, and F is a technical term flag.

20 is a display device such as a CRT display; 21 is a source text display area that displays the original text to be translated; 22 is a translated text display area that displays the translated text of the translation result; and 23 is a source language/translation support that displays correspondence information between the source language and the target word. Represents the display area.

In the present invention, in the translation dictionary 19, a technical terminology flag F is prepared for each original language to provide an overview.

Information indicating whether each word in the original language is a technical term can be registered by turning on/off this technical term flag F. Please include information indicating whether it is a technical term or not.
Instead of flags, identification information such as numerical values may be provided, and this technical term identification information may be information that can identify a plurality of types of technical terms for each field.

The processing units other than the technical term extraction processing unit 13 and the original language translation correspondence display processing unit 17 are substantially the same as those in the prior art.

The technical term extraction processing unit 13 sets technical term flags I? , and performs processing to identify whether or not it is a technical term and extract the technical term.

The source language/translation correspondence display processing unit 17 is configured to display information when translation fails due to an analysis error by the source text syntax analysis unit 14 or a translation generation error by the translation generation unit 15.

Regarding the technical terms extracted by the technical term extraction processing unit 13 for the original text, the correspondence information between the original language and its translation is edited.
3, a process of displaying the correspondence information is performed.

[Effect]

The original language translation for the original text that failed to be translated.

Only technical terms registered in the translation dictionary 19 in advance are displayed in the source language/translation correspondence display area 23, so when editing after translation, you can refer to the original language/translation correspondence information and quickly and easily create a translation. .

You will be able to input.

In this example, the translated text is output to the translated text display area 22, but the input or correction of the translated text in post-editing work is also performed by inputting characters to the translated text display area 22.

When a translation fails, only the most necessary word information that has been registered in advance is selectively displayed in the original language/translation display area 23, so that an easy-to-read display is possible. However, it is sufficient that a large display area is not required. In addition, the display area 23 corresponding to the original language and translation
Scroll display can also be performed.

〔Example〕

FIG. 2 is a processing flow of an embodiment of the present invention, FIG. 3 is an example of the structure of word data in the original language dictionary used in the embodiment of the present invention, FIG. 4 is an example of translation according to an embodiment of the present invention, and FIG. The figure shows an example of a bilingual editing screen according to an embodiment of the present invention.

The flow of the machine translation process according to the present invention is as shown in FIG. 2, for example. ■~[Phase] in the following description corresponds to processing ■~[Phase] shown in FIG.

■ Input the original text to be translated one sentence at a time.

■ Perform morphological analysis on the original text, divide the original text into word levels, and output conceptual symbols for associating the notation of the word with the corresponding translated word. Also,
A technical term flag indicating whether each source language word is a technical term is also output.

■ Check whether each source language word is a technical term using the technical term flag.

■ If it is a technical term, perform the secondary process ■.

If the term is not a technical term, skip process ■.

■ Storing the original language notations and conceptual symbols determined to be technical terms in a predetermined work area.

■ Determine whether all original words in the original text have been processed. If there are any unprocessed words, return to process ■ and repeat the extraction of technical terms.

■ Dependencies in the original language analyze relationships. If this syntax analysis fails, the process moves to [phase].

■ Generate a translated sentence based on the syntax analysis results. If translation generation fails, the process moves to the process [phase].

■ Output the translation. After that, continue translating the first original text.

[Phase] If the translation fails, use the translation dictionary 19B shown in Figure 1 based on the conceptual symbols of the technical terms memorized in process
, and match the source words with the translated words. The original language translation correspondence information is displayed on one display device 20.

Each word data of the original language in the original language dictionary 19A shown in FIG. 1 has a structure as shown in FIG.

Each word data consists of an original language notation 30, a conceptual symbol 31 internally determined from the meaning of the seven original words, grammatical attribute information 32 including two parts of speech, and various management information 33. In this management information 33, a technical term flag F is provided.

Next, a specific example of translation according to an embodiment of the present invention will be explained.

This will be explained according to FIG.

[Ibuana is a large lizard that lives in the Galapagos Islands.]
Suppose that we input the original sentence (a) and translate it.

In the original text morphological analysis, as shown in Figure 4 (b).

Divide the original text into words and associate the original language notation with a conceptual symbol for each necessary word. Here, whether or not the term is a technical term is also memorized.

Next, the morphological analysis result is subjected to syntactic analysis and translation generation, and if the translation is successful, the translation (C) is output.

If the translation fails, the translation dictionary 19B is searched for the conceptual symbol B for the technical term as shown in FIG. 4(d). Then, the translation notations corresponding to the conceptual symbol A and the conceptual symbol B, respectively, "iguana"
"and °'Galapagos l5lands",
Obtained from translation dictionary 19B.

Next, these original words and translated words are associated with each other and displayed on the display device 20.
to be displayed.

FIG. 5 shows an example of the bilingual editing screen. The bilingual editing screen consists of an original text display area 21, a translated text display area 22, and a source language/translation correspondence display area 23. 50 is a cursor that indicates a position on the screen.

Here, the Japanese sentence input into the original text display area 21 is translated into English. If the translation of each sentence is successful, the translated sentence is displayed in the translated sentence display area 22. The sentence number is ro
If the translation fails, as in the case of sentence 02J, the translated sentence will not be displayed, and information regarding the technical terms in the original sentence will be saved.

When the cursor 50 is placed on a Japanese sentence for which translation has failed, the original language translation correspondence information of the technical term in the Japanese sentence at the cursor position is displayed in the original language translation correspondence display area 23.

The operator refers to this technical term information and selects the translation display area 22.
By performing input operations similar to those used in a word processor, the correct translation can be entered.

In addition, if there is a Japanese sentence that has failed in translation, place the cursor 50 on that sentence and automatically display the source language translation correspondence information for the technical terms in that sentence in the source language translation correspondence display area 23, and input the correct translation. It is now possible to do this.

[Effects of the Invention] As explained above, according to the present invention, in translation work using a machine translation system, post-editing work such as searching for technical terms when translation fails can be significantly reduced, and the work can be done efficiently. You will be able to proceed with translation work 1] 2.

[Brief explanation of drawings]

FIG. 1 shows a configuration example of the present invention. FIG. 2 is a processing flow of an embodiment of the present invention. FIG. 3 is an example of the structure of word data in an original language dictionary used in an embodiment of the present invention. FIG. 4 is a side beam of translation according to an embodiment of the present invention. An example of a bilingual translation editing screen according to an embodiment of the invention is shown. In the figure, 10 is a processing device, 11 is an original text human resource section, 12 is an original text morphological analysis section, 13 is a technical term extraction processing section, 14 is an original text syntax analysis section, and 15 is a translated text generation section. 16 is a translation output unit; 17 is a display processing unit corresponding to original language and translation; 1
8 is an input device, 19 is a translation dictionary, 19A is an original language dictionary,
Reference numeral 19B represents a translation dictionary, F represents a technical term flag, 20 represents a display device, 21 represents an original text display area, 22 represents a translated text display area, and 23 represents an original language/translation correspondence display area.

Claims (1)

[Claims] In a machine translation system that performs translation using a computer, information (F) indicating whether each original language is a technical term or not.
A translation dictionary (19) that stores dictionary information to which is added, and a translation dictionary (19) that identifies whether or not each original word in the original text to be translated is a technical term based on the search results of the translation dictionary. A technical term extraction processing unit (13) for extracting technical terms, and a source language/translation correspondence display process for displaying correspondence information between the original language of the technical term extracted by the technical term extraction processing unit and its translation when translation fails. (17).