JP2726543B2 - Translation method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention receives a source language sentence represented as a sequence of character codes, translates the source language sentence by a transfer method, and obtains a sequence of character codes. The present invention relates to a translation method in a translation device that outputs a sentence in a target language expressed as “?

[Conventional technology]

FIG. 5 is a block diagram showing a main configuration of a translation device using a conventional translation method. In FIG. 5, reference numeral 1 denotes an entire translation apparatus, and 2 an analysis for analyzing a modification relationship between words in a sentence from a sentence in a source language input as a sequence of character codes and creating a structure representing the meaning of the sentence. Module 4 is a conversion module that converts the structure created by analysis module 2 into a structure for the target language, and 6 is a generation module that generates and outputs a sentence in the target language as a character code sequence from the structure for the target language. is there.

FIG. 6 is a flowchart for explaining a schematic operation of the conventional translation system, FIG. 7 is a flowchart for explaining an operation of extracting a classifier in the analysis processing of step 7 in FIG. The figure is a flowchart for explaining the operation of the processing of the classifier in the conversion processing in step 8 in FIG.

Next, the operation of this conventional example will be described with reference to FIGS. When the source language sentence is input to the translation device 1 shown in FIG. 5, the analysis processing of step 7 shown in FIG. 6 is executed. This analysis processing is performed by the analysis module 2. The process of cutting out the classifier in this analysis process is the seventh process.
As shown in the figure, one code is followed by one code.
When the code of an individual or a classifier is found, it is only used as the classifier and the classifier is not considered, and the case where the classifier continues after that is not considered. That is, in FIG. 7, the leading (next) element X is extracted from the character code list (step 10), and the element X
Is a classifier code (step 11), the element X
Is provided with information that it is a classifier (step 12).
In addition to such processing, the analysis module 2 analyzes the modification relationship between the classifier and another word or another word to create a structure representing the meaning of the input sentence.

When the structure representing the meaning of the sentence in the source language is created by the analysis module 2, the conversion process of step 8 shown in FIG. 6 is executed. This conversion processing is performed by the conversion module 4
Is responsible for. In the conversion process, the processing of the classifier is performed on the assumption that there is only one classifier after the number as shown in the flowchart of FIG. In FIG. 8, it is determined whether or not the classifier is translated in the target language (step 17). If it is translated, whether or not the classifier should be plural in the target language is determined. Is determined (step 19). If it is to be pluralized, the classifier has information indicating that the plural is formed in the target language (step 20).
If NO in step 17, the classifier is deleted (step 18). The conversion module 4 performs a conversion process related to other words in addition to the above process, and converts the structure of the source language sentence into a structure for the target language.

When the structure for the target language is created by the conversion module 4, the generation module 6 generates a sentence of the target language from the structure while using information in the structure itself.

[Problems to be solved by the invention]

As described above, in the conventional translation method, in the analysis of the source language sentence, the classifier is processed as being not continuous, so for example, in a Japanese expression such as "10 man months", "human"
However, there is a problem in that when any number of classifiers is combined or repeated, it cannot be correctly analyzed, for example, is correctly analyzed as a classifier, but the subsequent "month" is not analyzed as a classifier. Also, when converting the structure of the source language sentence to the structure of the target language sentence, it is assumed that the number of classifiers is only one, so if a number is followed by a plurality of classifiers, However, there have been problems such as the inability to delete classifiers that should not be translated in the target language, and the inability to set information indicating plural forms. As a result, there is a problem that a sentence in which any number of classifiers are combined or repeated cannot be correctly translated.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has as its object to obtain a translation system that can correctly translate even when any number of classifiers are combined or repeated.

[Means for solving the problem]

In the translation system according to the present invention, the analysis module 2 includes a classifier analysis module 3 that analyzes a sequence of classifiers, and the conversion module 4 includes a classifier conversion module 5 that converts a sequence of classifiers into a structure for a target language. , When the source language sentence is entered as a sequence of character codes,
When the analysis module 2 determines that the classifiers in the input sentence are continuous, the classifier analysis module 3 analyzes the classifier as a continuous classifier, analyzes the modification relationship between the classifier and other words, and The conversion module 4 creates a structure representing the semantic content of the language sentence, and the conversion module 4 converts the classifier analyzed as a continuous classifier into a structure for the target language by the classifier conversion module 5.

[Action]

In the analysis module 2, when it is determined that the classifiers in the input sentence are continuous, the classifier is analyzed as a continuous classifier by the classifier analysis module 3, and the modification relation between the classifier and another word is analyzed. A structure representing the semantic content of the source language sentence is created. And conversion module 4
In, the classifier which is obtained by analyzing the classifier as a continuous classifier is converted by the classifier conversion module 5 into a structure for the target language.

〔Example〕

FIG. 1 is a block diagram showing a main configuration of a translation apparatus using a translation system according to one embodiment of the present invention. In FIG. 1, components corresponding to those shown in FIG. 5 are given the same reference numerals, and descriptions thereof will be omitted. In FIG.
Reference numeral 3 denotes a classifier analysis module provided in the analysis module 2 for analyzing a sequence of classifiers, and reference numeral 5 denotes a classifier conversion module provided in the conversion module 4 for converting the sequence of classifiers into a structure for a target language.

FIG. 2 is a flowchart for explaining the overall schematic operation of this embodiment, FIG. 3 is a flowchart for explaining the processing of the classifier in the analysis processing of step 7 shown in FIG. 2, and FIG. 5 is a flowchart for explaining processing of a classifier in the conversion processing of step 8 shown in FIG.

Next, the operation of this embodiment will be described with reference to FIGS. When the source language sentence is input to the translation device 1 as a character code string, the analysis processing of step 7 shown in FIG. In this analysis processing, the classifier analysis module 3 is executed to extract the classifier. That is, in FIG. 3, the leading (next) element X is extracted from the list of character codes (step
10) If the element X is a code of a classifier (step 11), the information that the element X is a classifier is given (step 12). Then, the next element Y in the list is extracted (step 13). If the element Y is a code of a classifier (step 14), the element Y is given information that it is a classifier that is continuous with the element X (step 15). ). After the first classifier is cut out as shown in FIG. 3, it is checked in steps 13 and 14 whether the classifiers are not continuous.
Are parsed as continuous classifiers. In addition to the processing described above, the analysis module 2 analyzes the modification relationship between the classifier and another word or another word, and creates a structure representing the semantic content of the source language sentence.

When the analysis module 2 creates a structure representing the semantic content of the source language sentence, the structure is passed to the conversion module 4 and the conversion process of step 8 shown in FIG. 2 is executed. The classifier conversion module 5 is executed for the processing of the classifier in this conversion processing. That is, in FIG. 4, the head (next) element X of the list of classifiers is extracted (step 16). Next, it is determined whether or not the element X is translated in the target language (step 17). If it is translated, it is determined whether or not the element X should be pluralized in the target language. It is determined (step 19). If it is to be pluralized, the element X is provided with information indicating that it is pluralized in the target language (step 20). NO in step 17
In this case, the element X is deleted (step 18).

As shown in the flow chart of FIG. 4, after checking whether the first classifier is translated in the target language and whether or not the first classifier is to be changed plural times, next, the classifier is successively shifted. If so, the same processing is repeated. The conversion module 4 performs a conversion process on other words in addition to the above-described process, and converts a structure representing the semantic content of the source language sentence into a structure for the target language. When the structure for the target language is created by the conversion module 4, the generation module 6 performs the generation processing in step 9 shown in FIG. That is, a sentence in the target language is generated and output as a sequence of character codes based on the structure for the target language and information in the structure itself.

〔The invention's effect〕

As described above, according to the present invention, when it is determined that the classifiers in the input sentence are continuous, the classifier is analyzed as a continuous classifier, and the modification relationship between the classifier and other words is analyzed. A structure representing the semantic content of the language sentence was created, and the above classifier was converted as a continuous classifier into a structure for the target language, so if any number of classifiers were combined or repeated This also has the effect of correctly translating the classifier.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a main configuration of a translation apparatus using a translation system according to one embodiment of the present invention, FIG. 2 is a flowchart for explaining the schematic operation of this embodiment, and FIG. FIG. 4 is a flowchart for explaining the processing of the classifier in the analysis processing of step 7 shown in FIG. 2, FIG. 4 is a flowchart for explaining the processing of the classifier in the conversion processing of step 8 shown in FIG. 2, and FIG. FIG. 6 is a block diagram showing a main part of a translation apparatus using the translation method of FIG. 6, FIG. 6 is a flowchart for explaining the schematic operation of this conventional example, and FIG. FIG. 8 is a flowchart for explaining the operation of the processing of the classifier in the conversion processing of step 8 shown in FIG. 1 ... Translation device, 2 ... Analysis module, 3 ... Auxiliary analysis module, 4 ... Conversion module, 5 ... Auxiliary conversion module, 6 ... Generation module.

Claims (1)

(57) [Claims] An analysis module for analyzing a modification relationship between words in a sentence from a sentence of a source language input as a character code string and creating a structure representing the meaning of the sentence, and an analysis module created by the analysis module. A translation module for converting the generated structure into a structure for the target language, and a generation module for generating and outputting a sentence of the target language as a character code string from the structure for the target language, A module is provided with a classifier analysis module for analyzing the sequence of classifiers, and the conversion module is provided with a classifier conversion module for converting the sequence of classifiers into a structure for the target language, and the source language sentence is input as a character code string. Then
In the analysis module, when it is determined that the classifiers in the input sentence are continuous, the classifier analysis module analyzes the classifier as a continuous classifier, analyzes the modification relationship between the classifier and other words, and A translation method comprising: creating a structure representing the semantic content of a language sentence; and converting, in the conversion module, a classifier in which the classifier is analyzed as a continuous classifier into a structure for a target language by the classifier conversion module.