Researches of Sentence Similarity Computation Method Based on the Enhanced Petri Net

In the field of natural language processing (NPL), sentence similarity computation has a wide application. The key of sentence similarity computation is to grasp the characteristics and the meaning of the sentence quickly and accurately. In this paper, we present a new and advanced sentence similarity computation model based on the enhanced Petri net, which is mainly for the Chinese sentence. The basic idea of this model is utilizing semantic properties to expand the Petri net and reconstruct Petri net to an opening structure. The deduction method of this model is equation of state, which make the judgment of sentence similarity closer to the process of human thought. Moreover, we have tested the sentence similarity computation methods and algorithms. It proves that this method covers most aspects of sentence similarity computation and has a superior performance during experiments.