The Research Of Public Opinion Information Mining System

With the rapid development of science and technology, the network has become the place where China's public to express public opinion, to discuss public affairs, to join in social and economic participation and to carry out supervision of important public platform. Following with this is the problem how can relevant government departments control the public opinion. how To the find hot spots in time, how to correctly lead the public opinion, become in dire need of resolved.Mining public opinion is coming into being Based on the above background, the traditional information mining technology include three ways: Internet content mining, Internet log mining, Internet hyperlink mining. With the growth of pluralism of information, information of public opinion against mining is often limited for areas and for specific themes, and a single mining engine (such as Page Rank, HITS) is imprecise at recall rates and precision rates, apparently can not satisfy the users of a particular topic needs more in-depth digging. Therefore, this paper will combine web content mining and hyperlinks mining, introduce text similarity technology to the PageRank algorithm, formed Cosine Similarity PageRank (CSP) algorithm. After several experiments, we found that CSP algorithm has a more significant result in optimization of the entire network's PageRank rates.In this paper, we used CSP algorithm as the core, constructed public opinion mining system including the mining module, the mining policy management module, the user management module, and the Fault Management module. The system's major job is finish public opinion mining, text classification, concept Clustering, the similarity analysis, automatic Digest, and optimize the result of the web system,mail system and other major media networks. After establish the theme, the system will optimizing the initial seed and get the resource, then get tabloid through text analysis technology, After that the documents will be classified to the concept of clustering, a concept of space and a neural network algorithms will be settled, so that it can provide users with the concept of space-based interface for the research and the circumstances surrounding the development process,The system provide the appropriate information platform for the departments to regulate public opinion timely and reasonable.