Sentence Similarity Computing Theory And Applied Research

Chinese sentence similarity computation is a fundamental and important work in the Chinese information processing. It directly determines the status of research and development of certain related fields. For example, in FAQ System, information retrieval, Paraphrase, EMBT etc, sentence similarity computation is a very critical issue, which is also a hotspot and difficulty that people have studied for a long time.In order to study Chinese sentence similarity computation, the paper studied the similarity computation of three levels:primitive, word and sentence. And for the current sentence similarity computation methods, the paper proposed an improved method for sentence similarity computation, and finally introduces the FAQ system as an instance of sentence similarity computation.The main works completed in this paper are as follows:1) The study of primitive similarity computation and word similarity computation. Because the primitive similarity computation is the basis of word similarity computation, and word similarity computation is based on sentence similarity computation. Therefore this paper studied the primitive similarity computation and the word similarity computation in detail based on the "HowNet".2) Proposed an improved method for sentence similarity computation. Under the premise of the above work, this paper studied sentence similarity computation, and proposed a new method, which takes varied sentence features into consideration to calculate the sentence similarity. The three features integrated are:the word form information, the semantic information and the syntactic information. And, the traditional methods for the word form similarity calculation and semantic similarity calculation have been improved in this paper in order to compute the sentence similarity more accurately.3) Introduced the FAQ system as an instance of sentence similarity computation. To illustrate the practical application of sentence similarity computation, this paper studied the FAQ system and the sentence similarity calculation method adopted in which.