The Research On Chinese Sentence Similarity Algorithm Based On HNC

In this thesis, similarity computing was studied and the emphasis is sentences similarity computation based on word similarity computation. Firstly, different methods of word similarity computation are analyzed, and then the HNC-based word similarity computation method is realized, which shows its good performance at present. Secondly, the method used to judge synonyms and antonyms which is based on HNC is proposed. It can simplify the process of word similarity computation, and we realize it on computer. Then, after word similarity computation, for the reason of previous methods didn't take sentence structure, the importance of words that they act in the sentences and the different role of the words in the sentences into consideration, a new method of calculating sentences similarity based on HNC semantic block is presented. Finally, the new sentences similarity computation method is used in subjective questions auto-check. Application practice shows that the proposed method in this paper is more close to people's judgment than current methods in the aspect of determining logic errors and understanding semantic meaning. Furthermore, this method is easier to achieve and operate.To be more specific, the main work and research results in this thesis are as following:(1) The methods of Chinese word similarity computation are analyzed, and the HNC-based word similarity computation method is realized, so we can use it in the computation of sentences similarity.(2)A method used to judge synonyms and antonyms which is based on HNC is proposed. Whether the word is a synonym or a antonym is determine with rules. Moreover, the introduction of semantics simplifies the calculation of the word similarity.(3) On the basis of word similarity computation, a new method of calculating sentences similarity based on HNC semantic block is present. The method takes full consideration of the importance and the role of each word in the sentence.(4) The sentences similarity computation method based on HNC semantic block is used to check the interpretation of terms. The comparison between people' manual checking and auto-check shows that the proposed method is more practical and effective.