Ranking Algorithm Based on User Behavior Model

PageRank algorithm is based on network link to calculate the importance of the web pages. It is successful for web search, but there are still many problems, such as web spam, invalid links etc. It can not be a good description of the user's behavior. In this paper, through the analysis of thousands of user's behavior data, a semi-Markov process method is proposed to simulate the user browsing behavior, and its stationary probability distribution can be used as the measure of page im-portance. At the taking the influence of web content and length into consideration and com-bining with traditional hyperlink analysis, the new web pages are reconsidered and the old web ones are ranked appropriately. The results show that degree of satisfaction by this method is increased by about24% compared with PageRank algorithm, which can be better to calculate the importance of web pages.