BBS Monitoring System Using User Modeling And Topic Detecting

The rapid development of information ear makes most people share news and viewpoints online. Internet forum, in other word, BBS, has changed the pattern and spread of public opinion in a way. However, because of the freedom, interactivity and openness of internet, BBS has been a high occurrence zone of sensitive information and advertisements. How to hold the directions of the opinions of BBS and construct a clean and positive communication platform has become a problem for us to solve. In this thesis, we study how to modeling users’interests in BBS and use these interests to monitoring users’behavior. We also study some topic detection algorithms and use one of them to detect burst events on BBS.User modeling is to find, extract and express users’ interests. Through the reasonable expression of users’ interests, BBS administrator can understand uses’ concerns and make a corresponding processing. In this thesis, we provide a user monitoring method based user interest model. With calculating the attention value of sensitive post for each user and acquiring their relationship model, we can define users’ sensitiveness and estimate their potential behavior.Burst event is defined as the content wide spread in a short time. In BBS, burst events can be hot topics spontaneously discussed by users. They are more likely to be advertisements which are submitted repeatedly. In our thesis, we innovatively use topic detection algorithm based on burst futures in BBS to find hot topics and advertisements.Experimental results show that user monitoring can prevent publishing of sensitive subjects. Post monitoring can find most of the advertisements in BBS. With the basis of these two algorithms, we design and implement a BBS monitoring system to help administrator analyze internet forum’s information deeply.