A Mining Model For Web Short Text Based On Decision-Theoretic Rough Sets

With the development of Internet technology, the number of Web short text is growing every day, such as online goods comments. The Web short text is less words, strong emotion tendentiousness and dissipated structure and so on. The analysis and research of the short text has become increasingly urgent. So, This thesis proposes a Mining Model for Web Short Text based on Decision-Theoretic Rough Sets. This model is oriented to high-dimensional sparse text data such as online reviews, news reviews, online social dynamic of the online goods, which are updated quickly and have high randomicity in their content and is strongly random with high-dimensional sparse text data. With the advantages of the inconsistent, Decision-Theoretic Rough Sets is playing an import role to deal with vague and uncertain data, and simplifies the informative decision table. The automatic oriented knowledge clustering algorithm based on Decision-Theoretic Rough Sets was used to analyze the text data of web. Because online comment text emotional tendency is obvious, this thesis puts forward the emotional tendency analysis algorithm based on HowNet dictionary to make the high-dimensional text data into low dimension data. It is puted forward the ordered binary group of vector space model based on sparse matrix compression and vector space model, and we gave the corresponding calculation method, greatly reducing the storage space and optimiz the computation algorithm.We builted the online goods review analysis system by the JAVA programming. After geted Taobao online comments, we make the analysis of emotional tendency, and construct decision information table of low dimension. This mining text model of web based on the Decision-Theoretic Rough Sets was realized.