SIMPLE SENTENCE CLASSIFYING SYSTEM BY SEMANTIC CONTENTS

摘要

PURPOSE:To facilitate the construction of large scale knowledge by generating a similar group aggregation constituted of simple sentences whose semantic contents are similar to each other. CONSTITUTION:In a similarity calculating part 1, the processing by a word level similarity calculating part 11 and a sentence level similarity calculating part 12 is executed with regard to all simple sentence groups of a sentence aggregation being an object, and a similarity matrix being a result thereof is stored in a storage part 122. A similar group generating part 2 consists of a start set generating part 21, a representative simple sentence generating part 22 and a simple sentence group editing part 23, and stores a new representative simple sentence aggregation formed by unifying representative simple sentences to one with regard to a star (an aggregation of a reference sentence and a sentence being in a relation of similarity exceeding a threshold value), in a storage part 232. Accordingly, the sentence aggregation being the object can be ground automatically into groups whose semantic contents are similar. In such a way, a new construction of input sentence analyzing knowledge of a natural language processing system and an analysis processing of an input sentence aggregation can be executed in a short period.