The Semantic Computing Model of Sentence Similarity Based on Chinese FrameNet

The semantic similarity computing among sentences is important in many fields of NLP. This paper proposes a sentence similarity theory based on Chinese Dependency Graph (CDG) from Chinese FrameNet (CFN). Firstly, the CFN project is introduced, and then the CDG model is established. To compute the semantic similarity between two sentences, this paper proposes a series of similarity models. The target word similarity, dependency graph similarity and external component similarity are computed separately between two sentences, then the similarity between two sentences is defined as the convex combination of the three similarities. Compared with other theories, the sentence similarity theory not only computes the semantic frame similarity but also the word similarities between two sentences. Compared with other methods, the algorithm based on CFN and CDG makes the precision of sentence similarity obtain distinct improvement.