| With the development of Web2.0, Social networking sites continue to extend its function. As the new Internet product in recent years, the emotion computing and analysis of Micro-blog has become an important research subject in computer science, linguistics, natural human psychology and other social computing. This paper first introduces fuzzy clustering and the application of sentiment analysis in Tencent QQ space which verified the operability of the algorithm in sentiment classification. Fuzzy clustering analysis should make the text data numerical firstly, that is computing the sentiment of the blog. According to the principle and technology of web crawler, we designed the data acquisition system specifically for Sina micro-blog and build the source database. The word similarity computing method based on HowNet emotional word set calculates the weight of emotional word of micro-blog, establishes micro-blog emotional lexicon. In this process, network emotional lexicon is built for the need for future. Based on the emotional lexicon, the paper analyzes the emotional intensity of micro-blog which including negative adverbs and adverbs. According to the relationship between phrases and sentences as well as the corresponding emotional value, we get the whole micro-blog’s emotional value. We screened50users’emotional value in10different periods as the original data matrix, dynamically classify and generate dynamic clustering figure by fuzzy clustering algorithm. Different threshold will get different classification. The best classification can be calculated by F-Statistics, then get each user’s emotion classification chart by SPSS so as to analyze the emotional changes of the users. The innovation of this paper lies in the indefinite of people’s emotion and classification analysis to the micro-blog by fuzzy clustering. Businesses can obtain different classification results according to different needs, so as to take corresponding measures. |
PDF Full Text Request