Study on text semantic similarity in information retrieval

Case grammar is a theory about sentence structure proposed by C.J.Fillmore. A sentence is composed of modal and proposition. A simple sentence is expressed of sentence stem and decorating case. Sentence stem is a kernel of the sentence. It includes infliction case, kernel verb case and acceptance case. This paper proposed a formalized model of the text semantic similarity and similarity algorithm based on the case grammar. The semantic meanings of a sentence stem decide the similarity of a sentence. To the similarity sentence, a vector is used for the decorating case to get similarity algorithm. In this way, it avoided the keyword barrier of the text retrieval based on keyword.