FUNCTIONAL VERB STRUCTURE DISCRIMINATION METHOD, FUNCTIONAL VERB STRUCTURE NORMALIZATION METHOD, THOSE DEVICES, AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To automatically discriminate which of a "functional verb structure" to be normalized and a "main verb structure" or " idiom structure" not to be normalized is pertinent to a predicate consisting of the combination of "operational noun+particle+functional verb" in an input document.SOLUTION: A topicalization invalidation level calculation part 61 calculates topicalization invalidation level from the appearance frequency in the prescribed corpus of the standard type and topicalization deformation sentence of a functional verb structure candidate, and an idiomatic expression calculation part 62 calculates a phrase level between "operational noun" and "particle+functional verb" in the functional verb structure candidate as an idiom expression level, and an automatic discrimination part 63 defines the topicalization invalidation level and idiom expression level as an identity, and discriminates which of the functional verb structure, main verb structure and idiom structure is pertinent to the functional verb structure candidate from the topicalization invalidity level and the idiom expression level of the functional verb structure candidate by using a classifier created by automatic learning on the basis of prescribed answer data related with the functional verb structure using the functional verb structure, main verb structure and idiom structure as a class.