JPS63237171A - Maintaining system for vocabulary dictionary of machine translation system - Google Patents

Abstract

PURPOSE:To improve a translation speed and translation efficiency by using a dictionary data in a dictionary file for a translation processing, as leaving it just as it is, without data-converting it. CONSTITUTION:In the dictionary file 8, a dictionary source record, which is covenient for a user to use as a dictionary information for one key word in order to prepare of correct it and is of a descriptive form, and a dictionary object record, which is convenient to refer to at the time of the translation processing and is of the description form, are set, and the respective records are given flags to discriminate which form they belong to respectively. The preparation and the correction of the dictionary is performed as displaying the dictionary source record on an outputting device 7. At that time, the dictionary object record, which has the same key word as the dictionary source record, is automatically corrected as well.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a terminology dictionary holding method for a system for machine translation of a first language into a second language, and in particular improves the efficiency of dictionary access and the integration of dictionary data management. This invention relates to a term dictionary retention method suitable for improving the terminology dictionary.

[Conventional technology]

The terminology dictionaries 11 (hereinafter simply referred to as dictionaries) used in the machine translation system include specialized term dictionaries, basic term dictionaries, user dictionaries, etc., and the types and numbers thereof are enormous. The dictionary file holds such a dictionary as one headword (key) per word and dictionary information record for the key.

Conventionally, the format of dictionary information records in dictionary files has generally been such that even users who have no knowledge of machine translation processing systems can create and modify them, and when performing translation, one machine uses the dictionary file. was accessed, read the dictionary information in the N format, converted it into a predetermined format and stored it on the main memory, and used it as dictionary data for translation processing.

In machine translation systems, the dictionary database created by the dictionary data writer and the dictionary data used during translation processing are described in 1, for example [Information Processing Society of Japan Papers Window J 1986].
It is discussed in the August issue as "Operation method of dictionary database in machine translation".

[Problem that the invention seeks to solve]

The above conventional technology is convenient when a user creates or modifies dictionary information in a dictionary file, but it does not take into account high-speed reference by a machine translation system, and the translation program does not process dictionary data access. spend time on
This was a factor that led to a decline in system processing efficiency.

The object of the present invention is in a machine translation system.

It is an object of the present invention to provide a term dictionary holding method for a functional translation system that enables an improvement in the translation processing speed of the system without impairing the convenience of users in creating and modifying dictionary information in dictionary files.

[Means for solving problems]

The above purpose is to create a record (hereinafter referred to as a record) in a convenient description format for a user to create or modify dictionary information for one headword in a dictionary file.

(referred to as a dictionary source record) and referenced during the translation process (referred to as a dictionary source record).
(hereinafter referred to as dictionary object records) in a descriptive format convenient for accessing), assign a flag to each record to identify which format it belongs to, and modify, for example, a dictionary source record. In this case, this can be achieved by providing a means for automatically correcting dictionary object records having the same headword.

[For production]

When a user creates or modifies a dictionary, the user displays the dictionary source record on a display device, etc. This dictionary source record is written in a format convenient for displaying and modifying the contents. Even users who have no knowledge of machine translation processing systems can easily create and modify dictionaries.

On the other hand, when the translation program actually executes the translation process, it refers to the dictionary object record and performs the translation process.

Dictionary object records are written in a format that is convenient for accessing records, thus speeding up translation processing. Further, for example, when a dictionary source record is modified, the corresponding dictionary object record is also automatically modified, so that discrepancies in the contents of both can be prevented.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention, in which the central processing unit (
Translation processing device body) 1. Input device 6, output device! f7,
It consists of a dictionary file 8, an original text file 9, and a translated text file 10. The central processing unit 1 has a translation program 2, a dictionary maintenance program 3. It is equipped with a dictionary record conversion program 4 and an access management program 5.

The translation program 2 inputs the original text from the Jfi sentence text file 9, performs word division, imports each word information from the dictionary file 8, executes translation according to predetermined translation rules, and saves the translated text to the translation text file 10. Output. The dictionary maintenance program 3 adds/deletes dictionary information, starts an editor for modifying dictionary information, displays the contents on the output device 7, etc. according to key input from the input device 6, etc. The dictionary record conversion program 4 adds dictionary information, etc. - Access management program 5 that converts the dictionary source record into a dictionary object record when deletion, modification, etc. are performed.
is program 2, 3.4 and input bag w16. output device 7,
Dictionary file 8. It manages the exchange of information between the original text file 9 and the translated test file 10.

Figure 2 shows the schematic structure of the dictionary file 8. The dictionary file 8 holds one heading (m<key) and a dictionary information record for each word. Two types of records 21 and 22 are provided. The record 21 is a dictionary source record, which has a convenient description format for displaying and modifying the contents. record 2
2 is a dictionary object record, which has a convenient description format for reference (access) by the translation program 2.

FIG. 3 is an example of the dictionary source record 21, and FIG. 4 is an example of the corresponding dictionary object record 22.

In this case, translation from English to Japanese is taken up as an example, and rLOVEJ is shown as a headword.

31 and 41 indicate headwords (keys), and 32
．． 42 indicates record information. Note that N is a noun and ■ is a verb. 33.43 is a record identification flag, S indicates a dictionary source record, and 0 indicates a dictionary object record.

The description of the dictionary source record in FIG. 3 is basically the same as that of a normal English-Japanese dictionary, and can be easily created and modified even by a user who has no knowledge of machine translation processing systems. On the other hand, the description of the dictionary object record in FIG. 4 is similar to the arrangement of records in a memory in a machine search system or the like.

When executing translation, the dictionary file 8 can be easily accessed.

FIG. 5 shows how the central processing unit 1 retrieves the headword rLO of the dictionary object record 22 from the dictionary file 8 when executing translation.
A flowchart for importing VEJ information is shown. Upon receiving the request from the translation program 2, the access management program 5 selects the identification flag 43 from the dictionary file 8.
is "0" and the key is rLOVEJ (step 501). If a record exists, the corresponding record is immediately read (step 50).
3) If the record does not exist, information that the word rLOVEJ is an unknown word that does not exist in the dictionary is returned to the translation program 2 (step 502).

FIG. 6 shows a flowchart when the central processing unit 1 corrects the dictionary contents in the dictionary file 8. Upon receiving the request from the dictionary maintenance program 3, the access management program 5 searches the dictionary file 8 for a dictionary source record whose identification flag 33 is "S" and whose key is "rLOVE" and outputs it to the output device 7 (step 601). , the corrector updates the corrected dictionary source record using the input device 6 while viewing the displayed content. After that, the format of the updated information is converted using the dictionary conversion program 4 (Step 602), and the key of the dictionary object record in the dictionary file 8 is checked (Step 603).
If the key "OvE" exists, update the rLOVEJ record in the dictionary object record with this converted content (step 604); if the key does not exist, rLOVE
Create a new record of "V\E" in the dictionary object record section and write the converted content to dictionary file 8 (
Step 605).

〔Effect of the invention〕

As explained above, according to the present invention, the translation system can
When you read the dictionary data in the dictionary file, you can use it directly for N translation processing without data conversion.
It is possible to improve translation speed and efficiency, and by managing dictionary source records and dictionary object records in one file, it prevents discrepancies in the content of both and improves data reliability. can get.

[Brief explanation of drawings]

FIG. 1 is an overall block diagram of an embodiment of the present invention. Figure 2 shows the schematic structure of the dictionary file in Figure 1, Figure 3 shows an example of the dictionary source record in Figure 2, and Figures 2 and 4 show an example of the dictionary object record in Figure 2!14. FIG. 5 is a processing flowchart for searching a dictionary file during translation processing, and FIG. 6 is a processing flowchart for correcting/updating records in the dictionary file. 1... Central processing unit, 2... Translation program, 3
...Dictionary maintenance program, 4.Dictionary record conversion program, 5.Access management program,
6... Input device, 7... Output device. 8...Dictionary file, 9...Original text file, 10...Translation text file, 21...
Dictionary resource record, 22...Dictionary object record. Figure 3 ■Work! 41 Figure 5 Collection Figure 6

Claims (1)

[Claims] (1) In a machine translation system that automatically translates a first language into a second language, the machine translation system is equipped with a dictionary file that uses a first language as a headword and holds dictionary information of a second language translation for the first language, and the dictionary file includes: For one headword, the dictionary information consists of a first record in a descriptive format that is convenient for users to create or modify, and a second record in a descriptive format that is convenient for reference during translation processing. A flag is attached to each of the first and second records to identify which format they belong to, and when the first record is corrected, the second record having the same headword is also automatically corrected. A terminology dictionary retention method for a machine translation system.