Research And Realization Of Chinese Subjective Sentence Recognition Method Based On Fuzzy Set

With the rapid development of the WEB2.0 technology, opinion mining has become a hot topic of natural language processing. As an important sub-task of opinion mining, the main purpose of subjectivity recognition is to recognize subjective opinions from objective sentence that contain the description of objective fact in user generated text. It has very important significance that subjectivity recognition can reduce the complexity and improve the performance of opinion mining system.According to the characteristics of fuzzy boundaries between Chinese subjective and objective, this article explores the fuzzy representation of subjective characteristics in different granularity under the framework of fuzzy set theory, then we proposed fuzzy classification model based on features of different granularity to solve the problem of Chinese subjectivity recognition. Specifically, the paper includes the following aspects:(1) Chinese subjective sentence recognition based on subjectivity density fuzzy sets. First, we use the odds ratio formula to extract subjective clues from tagged corpus. Then, calculate the subjectivity density of Chinese sentences based on subjective clues. Finally, we build corresponding triangular membership functions based on subjective density for Chinese subjective and objective sentences and implement a Chinese subjective sentence classifier based on the fuzzy sets. Experimental results show that the proposed classifier based on fuzzy sets can identify the boundaries of subjective density between subjective sentences and objective sentences more exactly and we achieved better results.(2) Chinese subjective sentence recognition based on subjective/objective voca bulary fuzzy sets. Firstly, we define two fuzzy sets corresponding the fuzzy conc epts of subjective and objective vocabulary and employ TF-IDF based on fuzzy statistical method to acquire the membership function from the training corpus. T hen, set up two fuzzy IF-THEN rules in the framework of fuzzy set theory and accordingly implement a fuzzy inference machine to recognize Chinese subjective sentence. Experimental results show that compared with traditional subjective se ntence recognition methods, the proposed method based on vocabulary fuzzy sets can recognize the subtle differences between Chinese subjective and objective sen tences more accurately, therefore it improve the Chinese subjective sentence reco gnition performance significantly.