Research And Implementation Of User Behavior Analysis System Based On Web Log

Since the birth of the Internet,with the continuous development of networktechnology,the network has gradually become the primary way people access information andresources.The Internet has brought great convience to people,but meanwhile,after decades ofdevelopment it has accumulated a mass of users behavioral data.Faced with such a valuableresource,how to dig out the potential,hidden, valuable information from the massive userbehavior data and use them effectively,has become a hot research topic.Based on the analysis of Web logs,this paper have digged out some potentialinformation of users’ behaviors. Thanks to the existing Web data mining model and thealgorithm, we propose an improved AprioriAll algorithm to find out the users frequentlyaccess paths, and proposed a clustering algorithm based on user navigation path andfrequency of access. By changing the candidate sequences connections we reduce thegeneration of candidate sequences and reduce unnecessary database scanning, the improvedAprioriAll algorithm reduces the time and space complexity of the original algorithm. Inorder to get the user’s cluster, we first use the access path to get the Web site’s similaritymatrix of user, after then we get an initial user clustering. At last we use the formula todetermine the strength of each user within the initial cluster.The low cohesion will beremoved from the current cluster, thereby refining the initial cluster. Because at the same timetaking into account the user’s access paths and access frequency, accuracy algorithm to acertain extent.Finally, through the java language to implement a simple Web log mining prototypesystem, the path set for Web logs to dig out the user on the basis of pre-treatment frequentlyaccessed and users with similar access clustering behavior. Experiments show that therealization of this study user behavior analysis system has better achieve the effect, efficiencyand accuracy of the algorithm has been verified.