Research On Algorithms And Application Of BBS-Based Public Sentiment Prediction System

BBS is becoming a platform through which more and more people express their views and put forward their claims, so it's important for understanding the relative situation of real-world to do mining on those information. Data mining on the BBS topics by category can predicts the potential business opportunities and public sentiment etc, and provides support for commercial activity and guiding the public sentiment.In this dissertation, the present status and trends on BBS-Based data mining are introduced, the subject crawler technology and the classical public sentiment prediction model are in-depth researched, and an efficient, scalable public sentiment prediction system is designed and implemented. The main works are as follows:(1) Several classical subject crawlers and the classifier prediction method are researched, and then an accurate BBS-based subject Crawler is implemented by improving the classifier prediction method. The feature of BBS is considered by the new crawler, so it can quickly grabbed in useful information without evaluation on the page content linked by anchors.(2) In this dissertation, the classical Influence Diffusion Model is in-depth researched. The acquired topics are sorted into an appropriate problem space, and then the behaviors of opinion leaders are analyzed by statistical methods, at last, with the introduction of the attenuation property in the topics influence, a new public sentiment prediction method which can quantitatively predict the potential risk degrees of BBS topics more accurately is proposed.(3) Above-mentioned public sentiment prediction method is proved to be feasible by simulation.(4) Based on Above-mentioned public sentiment prediction method, a BBS-based public sentiment prediction system is established, and then the workflow and the main modules of the system are introduced in detail.