JPH08137884A - Proofreading method for japanese sentence - Google Patents

Abstract

PURPOSE: To securely and easily proofread a misused word and an unregistered word in Japanese processing. CONSTITUTION: The unregistered word is extracted and corrected by taking a morpheme analysis of the original sentence (S1) and the misused word is extracted by referring to a misused word dictionary for morpheme data of the corrected unregistered word (S2); and the extracted misused word is pointed out and a list of correction candidates is displayed (S3). When the misused word is pointed out, a terminal which seems correct is selected out of the correction candidates to proofread the misused words and unregistered word through correction (S4) and its confirmation (S5). A syntax analysis of the morpheme data obtained by the morpheme analysis is taken (S6), a syntax analysis of morpheme data obtained by correcting the misused word is taken (S7), and when both the syntax analytic results have a difference in sentence meaning, an error of the correction is decided and recorrection is pointed out.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Japanese language processing system, and more particularly to a proofreading method for erroneous terms and unregistered words in documents.

[0002]

2. Description of the Related Art Japanese language processing using computers such as word processors, machine translations, document databases, and hypertexts has been put into practical use.

In natural language analysis for this purpose, a sentence to be analyzed is first divided into morpheme units (minimum units of word structure), and morpheme analysis is performed to clarify the properties of each morpheme. After this, a syntactic analysis that analyzes from the syntactic rules of natural language, and then a semantic analysis that removes ambiguity and vagueness,
Perform contextual analysis.

In the syntactic analysis, it is determined whether or not the morphologically analyzed sentence is a correct sentence by using a grammar, and when the sentence is correct, a tree structure (a parse tree) is obtained as the syntactic analysis result.

In this syntactic analysis process, since attention is paid only to the grammatical conformity, syntactic ambiguity occurs and many syntactic trees are generated. A semantic analysis process is performed to select the correct parse tree from among these.

In the semantic analysis process, not only the grammatical category of a word (corresponding to a part of speech) but also its semantic information is used.

In such Japanese language processing, a function of extracting and correcting erroneous terms and unregistered words is provided as one function of the document proofreading support system. This function is performed by the procedure shown in FIG.

(S1) Morphological analysis is performed on the original sentence to be processed in Japanese, and unregistered words are extracted. This unregistered word is directly corrected by the operator.

(S2) The morpheme data which is the morpheme analysis result is read, and the erroneous term is extracted by referring to the dictionary. If there is an erroneous term in the corrected unregistered word, it is extracted.

(S3) The extracted erroneous term is pointed out to the operator on the display screen, and a list of correction candidates is displayed.

(S4) In response to this indication, when the operator selects a term that seems to be correct from the correction candidates, the term is replaced.

(S5) When the operator's confirmation is obtained by replacing the terms, the next erroneous term is extracted, and when all the erroneous terms are corrected and confirmed, the completion is completed.

[0013]

Since the conventional method is carried out as a manual calibration work, it has the following problems.

(1) If there are many correction candidates, there is a possibility that the wrong candidate will be selected.

(2) If there is no correction candidate, it may depend largely on the operator's knowledge and an appropriate correction may not be possible.

An object of the present invention is to provide a proofreading method that surely facilitates proofreading of erroneous terms and unregistered words.

[0017]

In order to solve the above problems, the present invention performs morphological analysis of Japanese sentences, adds unregistered words to the analysis result data, extracts erroneous words, and corrects them by an operator. In the method of proofreading incorrect terms / unregistered words, the syntactic analysis is performed on the morpheme data obtained by the morphological analysis, the syntactic analysis is performed on the morpheme data obtained by correcting the incorrect terms, and both syntactic analysis results are converted into sentences. The feature is to point out the re-correction when there is a discrepancy in intention.

[0018]

The correct correction is obtained by checking whether or not the correction is appropriate based on whether or not the morpheme data is different from the morpheme data in which the erroneous term is corrected.

The operator's burden is reduced by automatically checking whether or not the correction is appropriate and by pointing out the correction again when the correction is not appropriate.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 is a processing procedure diagram showing an embodiment of the present invention. 2 is different from FIG. 2 in that processing elements S6 to S
8 is provided.

In the syntactic analysis S6, syntactic analysis is performed using the morpheme data which is the morpheme analysis result. This processing uses the parsing function of the conventional Japanese language processing system.

In the syntactic analysis S7, syntactic analysis is performed using the morpheme data corrected by the process S4. This processing is also the same as the syntax analysis S6.

The textual analysis S8 is a syntactic analysis S6. The analysis results of S7 are compared, and it is determined by semantic / context analysis whether or not a semantic change has occurred in the syntax. This processing uses the semantic / context analysis function of the conventional Japanese language processing system.

When a difference is made in the meaning of the sentence in the judgment of the meaning analysis S8, that is, as a wrong term (including the added unregistered word), a warning is issued and an indication is made by the indication processing S3, and the correction processing S4 is performed again. Return to, and perform a textual comparison by correcting again.

If there is no difference in the meaning of the sentence determination, the process proceeds to the next incorrect term determination S2 as an appropriate correction.

Therefore, according to the present embodiment, when the syntactic analysis results of the morpheme data obtained by the morpheme analysis and the morpheme data obtained by correcting the erroneous term are different from each other in the sense of meaning, the correction is again performed as an inappropriate correction. Point out the fix.

As a result, it is possible to immediately correct erroneous words and unregistered words due to warnings and indications of mistakes in the meaning of the sentence, and it is possible to always obtain a high composition quality without being influenced by the knowledge and assumptions of the operator. Can be kept.

Further, when the incorrect term / unregistered term is corrected,
In the past, it was necessary to read and compare that the corrected sentence and the original sentence did not have a literary difference, and a huge amount of correction time and work load were imposed. Since the sentence meaning is automatically evaluated, the correction time and the work load can be greatly reduced.

[0029]

As described above, according to the present invention, the syntactic analysis is performed on the morpheme data obtained by the morpheme analysis, and the syntactic analysis is performed on the morpheme data obtained by correcting the erroneous term,
Since corrections are pointed out again when there is a literal difference between both parsing results, there is an effect that it will be a reliable calibration, the calibration quality will be improved, and the operator's work time and burden will be greatly reduced. .

[Brief description of drawings]

FIG. 1 is a processing procedure diagram showing an embodiment of the present invention.

FIG. 2 is a diagram of a conventional processing procedure.

Claims (1)

[Claims] 1. A method of morphologically analyzing a Japanese sentence, adding an unregistered word to the analysis result data, extracting an erroneous term, and correcting it by an operator, thereby correcting the erroneous term / unregistered word. Perform syntactic analysis on the analyzed morpheme data, perform syntactic analysis on the morpheme data in which the incorrect term is corrected, and point out recorrection when the syntactic analysis results are literally different. A proofreading method for characteristic Japanese sentences.