Research On Hot Topic Discovery Model In Web Enviornment Based On User Browsing Behavior Analysing

Along with the rapid development of Internet, network news has become an important approach to gain information, and its influence in readers has also increased consumedly. Meantime, network news sources are numerous, with totally different angle and position, which leads to information overload problem. Therefore, how to handle the large amount of network news and discover hot topics on web rapidly is very useful in many fields, not only for government to adjust public opinion tendency and keep social stability, but also for readers to find out interested news in related topics timely.This paper stands on the basic of User Browsing Behavior and Hot Topic thesis and takes network news flow as its research object. TDT method is used for clustering reports according to topics and a Network Hot Topic Discovery Model framework based on User Browsing Behavior is proposed. The main work in the research includes the following elements:1. Quantitative description of Topic Attention Degree According to TF*PDF algorithm, a quantitative description of Topic Attention Degree is presented, which is used to describe the attention degree of a topic in a web site within a period. Hot topic is defined as a topic with both high media attention degree and high user attention degree at the same time. User attention degree is proposed based on hot topic thesis and User Browsing Behavior analyzing.2. Description and introduction of User Browsing BehaviorIn Psychics, psychology status is presented by some special behavior information. As a result, user browsing behavior is collected stealthily and formed user attention degree.Two key problems are solved here: user browsing behavior information which is fit for describing user attention degree is picked up and combined to describe the news' user attention degree quantitatively.3. How to describe the life cycle of hot topicsBased on decay theory and the idea of stock index, topic Index curve is introduced to describe the life cycle of hot topics, which is useful to analyze and forecast the development of topics on web.Hot topics on a web site can be detected automatically by this model within a time period by analyzing the news flow and the user browsing behavior. Hot topic index of each topic can also be obtained from this model. It is proved by experiment and readers' survey feedback that this model is useful in hot topic detection and user browsing behavior is important in hot topic discovery and Hot Topic Index curve could describe the life cycle of a hot topic.