Researcn Of Micro-Blog Sentiment Analysis Model Based On Combination Of Sentiment Dictionary And Rules

Since the21st century, China’s Internet industry has been booming development, netizens are increasing year by year. On the Internet in recent years, micro-blog is increasingly popular pastime, on the political and business elite, down to the ordinary people are fun, micro-blog has gradually become many people’s lives essential element. Sina micro-blog platform produces hundreds of millions of micro-blog to share content, and disseminate information every day, behinded the huge amount of users and the amount of data is accompanied by hidden commercial, social and many other value.Research on the micro-blog sentiment analysis is the process to explore micro-blog’s potential business, social and other value, research of micro-blog sentiment analysis can be applied to detect and monitor public opinion, forecast information, evaluate and improve product. In-depth study of micro-blog content, and access to micro-blog emotional tendencies are very necessary.Currently micro-blog emotional polarity classification method has shortcomings of significantly lower accuracy rate, dependent domain knowledge, consider the relationship between sentences and in sentence, our study hopes to find a way to make the classification accuracy can be improved, universal performance of the method has been strengthened. Based on this starting point, the paper studies the micro-blog sentiment analysis method combined emotion dictionary and rules, the main contents include the following two parts:(1) By constructing emotional dictionary, access semantic rules, as emotional word for center, summed up the six kinds of emotional word combinations, taking into account between the emotional words, negative words, and interaction adverbs, combining emotion dictionary and rules, we use micro-blog emotional value clause, sentence emotional value calculation method, achieve the micro-blog emotional polarity classification ultimately. Experiments show, the proposed method is better than the emotional symbol discrimination method, emotional dictionary discrimination method, SVM discrimination method in micro-blog emotional polarity classification results.(2) Based on (1), we study the impact of the conjunction of emotional expression micro-blog, from four kinds of general usage of the conjunction, considering the micro-blog’s relationship in sentences, the relationship between sentences, we introduce of the conjunction weight factor to improve (1) emotional value micro-blog clause, sentence emotional value calculation method, enhance micro-blog emotional polarity classification results. Experiments show, the classification results of approach considering the conjunction has been improved than the previous method. Overall the experimental comparison of the proposed authentication method is not dependent on knowledge of the field, pervasive strong, high accuracy.