Research Of User Preferred Browsing Paths Based On Web Log

With the development of Internet technology,storage technology and computing power,more and more business activities are trading through the network platform.Websites have accumulated a large amount of click-stream data in their daily operations,which provide us with a good opportunity to analyze and mine valuable information.This paper uses the data to mine user's preferred browsing paths,which can refactor the links between pages,optimizing the websites to improve user's access experience and achieving better market competitiveness.Frequent-preferred path algorithm,page value-jump preference algorithm and support-preference algorithm are best algorithm nowadays.Frequent-preferred path algorithm takes user's accessing frequency as the main reference factors,ignoring user's browsing interest.Although page value-jump preference algorithm takes user's browsing interest as the main reference factors,but it is strict to the quality of data in the practical application.Although support-preference algorithm takes user's browsing interest as the main reference factors and it's not so rigorous to the data,but the main path of website doesn't being filtering out from the browsing paths which have been minied.Therefore,this paper puts forward a weight matrix and the effective-preference algorithm to improve the support-preference algorithm,which can weaken the influence of the main path from mining results.In this paper the main research work are as follows:(1)Summary the content and development of web mining,and obtain the practical significance of mining user's preferred browsing paths by analyzing the problems which have existed in the website.This paper sums up good algorithms recently of mining user's preferred browsing paths,which the main problem is that they are difficult to balance accuracy,complexity and efficiency.(2)In order to solve the problem of original web logs which can't be used directly to mine user's preferred browsing paths,this paper uses the data preprocessing method to deal with web logs.This method delete the original data which attributes and records have nothing to do with mining purposes,and session identification processing the data to find out the order of user's browsing.(3)On the critical problem of mining user's preferred browsing paths,we present a weight matrix and the effective-preference by weighting the topological structure of the website to improve the support-preference algorithm of mining user's preference browsing path,which can avoid mine the main path from the topological structure of the website.It proves that the improved algorithm is effective and feasible by an example analysis,and it is better to reflect the user's real browsing interest than the former algorithm.(4)Using the web log of the Sogou to experiment,the result shows that the improved algorithm has filtered out the main path and has higher accuracy than the former.However,the improved algorithm spent more time to execute than the former because of the weight matrix.So the improved algorithm exchanges higher accuracy with more time.