Text sentiment classification and discrimination is to dig through the automatic division of comments positive or negative views of customer feedback, an" />
Research on Computation of Chinese Text Sentiment Classification of Multi-granularity
|Course||Computer Software and Theory|
|Keywords||Text sentiment classification Emotional tendencies Text Dependency grammar Sentence polarity|
Text sentiment classification' dissertation">Text sentiment classification and discrimination is to dig through the automatic division of comments positive or negative views of customer feedback, and to appraise the identification and classification of a product. Text related to emotional classification generally vocabulary, sentences, paragraphs and chapters. The complexity of the Chinese language tend to make the vocabulary, sentence and text emotion detection research is facing a lot of difficulties. In text sentiment classification computing research methods of the various levels of granularity, and not enough detailed knowledge and understanding involving to some linguistics emotional factors, resulting in some study emotion discrimination appears subjective judgment inconsistent. Therefore, this article first in Hownet based on the calculation method of lexical semantic tendency to be improved, expanded a variety of factors affect the emotional and linguistic knowledge. Finally, the Chinese sentence subject extraction and polarity discrimination detailed improvements, and make use of with the rule matching algorithm for the calculation of the sentence emotional tendencies. The following are the main research content of this paper: (1) Text theoretical premise of the calculation method based on the Chinese tendency to emotional vocabulary. Specific solution the HowNet concept of justice in the original description of dislocation and the emergence of the concept of ambiguity problem, this part of the work is conducive to the subsequent sentences emotional tendencies discriminant. (2) the sentence sentiment Influencing Factors to consider and quantitative analysis. First, the negative words to expand, and with the vocabulary specific modification polarity quantitative semantic orientation, further study Chinese sentence negation shared, as well as the negative polarity of the question of the transfer of the comparative sentence emotional, because these factors tend to affect emotional tendencies of complex sentence structure. Secondly, a detailed analysis of the exclamatory, respectively, from interjection match rules start from the linguistic point of view, be all sorts of possible exclamatory forms and expressions of sentiment analysis. (3) improve the sentence polarity discrimination no longer rely solely on the syntactic structure of relations in the dependency structure to transfer the polarity value, but according to the definition of the dependency structure between the first sentence subjects extraction, and then consider the dependency grammar distance and modification of the the word polarity of the sentence polarity. Sentence negation rules match, making the sentence level polarity discrimination more accurate. Secondly, in the vocabulary, sentence sentiment tendency to study the basis of the emotional tendencies of the Chinese version of the analysis and calculation and design applications. Finally, the proposed vocabulary, sentence sentiment tendency of the method used, the system of evaluation and other Text-based research methods, the experiment results, the results show that the effectiveness of the proposed method, and the emotional classification accurate and precision rate increased.