The Design And Implementation Of Multi-features Combination In Sentence Similarity Computation

In the research field of natural language processing,the calculation of sentence similarity is a very basic and critical research work and its accuracy directly affects the progress of some other related research fields such as: Chinese automatic question answering system,sentence similarity is used for matching user input and the questions in the corpus,the corresponding answers returned to the user according to the matching result;in the automatic summarization system,sentence similarity is used for eliminating the similar sentences and extracting the main sentences;in the example-based machine translation,sentence similarity is used for matching the bilingual alignment sentences in the corpus,the translation of the sentence can be obtained according to the example translation results.The characteristics of the sentence similarity calculation method is studied,including word form similarity,word sequence similarity,structure similarity,word POS similarity,syntactic similarity and semantic similarity.The similarity of structure is improved by using a simplified version of the structural similarity calculation method;Due to the different importance of different parts of the dependency relationship,the two dependency relational five tuple similarity is presented.Based on the multiple features of sentence similarity,a method for calculating sentence similarity is proposed.Through two experimental results,we can see that the similarity calculation method proposed in this paper can meet the actual demand,and the similarity value is consistent to the expert evaluation.Finally,the full text is summarized,and the future development direction of sentence similarity is prospected.