JPS63138465A - Analyzing device for syntax structure - Google Patents

Abstract

PURPOSE:To improve the working efficiency with analysis of syntax structures by applying only a fixed syntax structure rule before analysis of the syntax structure to decrease the number of morphemes of an input sentence and reducing the analysis load of syntax structure at the next stage. CONSTITUTION:A syntax structure analyzing device contains a morpheme analyzing means 12 for input Japanese word sentences, a fixed rule deciding means 16 which decides whether just a single syntax rule is applicable to a part of the morpheme analysis result or not, a fixed rule application means 17 which uses the output of the means 16 to apply the syntax structure rule to the morpheme analysis result, and a syntax structure analyzing device 19 which applies the syntax structure analysis to a morpheme string undergone application of a fixed rule. The syntax structure rule is applied to an area where an applicable syntax structure rule can be fixed among those morpheme analysis results. Thus the number of morphemes are successively decreased and the result of this reduction is used as the input of a syntax structure analyzing part. In such a way, the analyzing time is shortened.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a parsing device for a natural language processing unit of a machine translation system, a database search system, or the like.

BACKGROUND OF THE INVENTION As a conventional method of analyzing natural language sentences to obtain syntactic structures, a structure as shown in FIG. 2 is known. In the figure, an input sentence 1 is read into a morphological analysis device 2 as a character string and decomposed into a string of morphemes having parts of speech and notation. At this time, the morpheme dictionary 3 is used to identify morphemes.

The output of the morphological analysis device 2 is supplied to the syntactic analysis device 5 as a morpheme sequence 4, the syntactic structure is analyzed using the syntactic rules 6, and the result is output as a syntax tree 7. One of the formats currently commonly used as syntax rule 6 is context-free grammar,
Alternatively, for each rule, there is an extended context-free grammar with some extended functions added. One of the advantages of using a context-free grammar is that a method (algorithm) is known that guarantees that parsing is completed in finite steps.

Problems to be Solved by the Invention However, conventional analysis methods have the problem that the time and storage capacity required for analysis rapidly increase as the number n of morphemes in an input sentence increases.

Ah! is said to be the most efficient! J-(Earley
), the time required for analysis is r
It is on the order of L3; some other methods are on the order of e%. The present invention solves the above-mentioned problems by applying syntactic rules in advance to locations where applicable syntactic rules are uniquely determined before performing syntactic analysis to reduce the number of morphemes in the input sentence. The purpose is to provide an efficient syntactic analysis method by using this as input to the syntax analysis section.

Means for Solving the Problems In order to achieve the above object, the present invention provides means for morphologically analyzing an input Japanese sentence and determining whether only one syntactic rule is applicable to a portion of the morphologically analyzed result. a deterministic rule discriminating means for applying a syntactic rule to a morphological analysis result using the output of the deterministic rule discriminating means, and a syntactic analysis device for parsing a morpheme sequence after applying the deterministic rule. It is something that

According to the above-mentioned structure, the present invention applies syntactic rules to sequentially reduce the number of morphemes in the portions of the morphological analysis results for which applicable syntactic rules can be determined, and inputs the results into the syntactic analysis unit. By doing so, the time required for analysis is greatly reduced.

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a parsing device according to an embodiment of the present invention. In the figure, nine Japanese sentences 11 written in solid letters, including kanji and kana, without any separation, are input to the morphological analysis device 12. The morpheme analysis device 12 refers to the morpheme dictionary 13, identifies morphemes in the sentence, checks whether adjacent morphemes can be connected, and converts the morpheme string 14 into
Output in order of validity. The confirmed rule discriminator 16 is
Search for context-free grammar rules that are uniquely applicable to the morpheme subsequence from the morpheme sequence 14 and context-free grammar rules 15, and if any exist, use these rules as definite syntax rules. It is output to the application device 17. The deterministic rule applying device 17 applies the deterministic syntax rules to the morpheme sequence 14, and as a result, the morpheme sequence 14 is modified by the deterministic syntactic rules and the number of syntactic elements is reduced. The definite rule discriminator 16 searches for a definite syntax rule again for the changed morpheme sequence, and if a definite syntax rule exists,
Application of the definite syntax rules again takes place in the definite rule application device 17.

This procedure is repeated until there are no more definite syntactic rules, and the length of the morpheme sequence is successively reduced.Finally, a shortened morpheme sequence 18 is output to the syntactic analysis device 19. The syntax analysis device 19 receives the shortened morpheme sequence 18 and parses it using the context-free grammar rules 15 to obtain a syntax tree 2o.

Effects of the Invention As described above, the present invention reduces the number of morphemes in an input sentence by applying only definite syntactic rules before parsing, and reduces the burden of subsequent syntactic analysis, thereby achieving efficient syntactic analysis. It can be done and the effects are great.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram of a Japanese language syntax analysis device according to an embodiment of the present invention, and FIG. 2 is a conceptual diagram of a conventional syntax analysis device. 11...Solid writing/Kanji-kana mixed syntax, 12...Morphological analyzer, 13...Morpheme dictionary, 14...Sequence of morphemes, 15...Context-free grammar rules, 16...Determined rules Discrimination device, 17... Determined rule application device, 18...
・Shortened morpheme sequence, 19... Syntactic analysis device, 20
...Syntax tree. Name of agent: Patent attorney Toshio Nakao and 1 other person No. 1
Figure 2

Claims (2)

[Claims] (1) A morpheme sequence after applying the syntactic rules, comprising a means for morphologically analyzing an input sentence and a means for applying the syntactic rules to a location where only one syntactic rule applicable to the morphologically analyzed sentence can be determined. A syntactic analysis device, characterized in that: is used as an input to a syntactic analysis means. (2) The syntactic analysis device according to claim 1, characterized in that a format of a context-free grammar or an extended context-free grammar is used as a syntax rule.