Based On Markov Chain Web Access Sequence Mining Algorithm

Network plays a more and more important role in people's daily life. There is an extensive research and application space to improve usability of the web site via users' behaviors.The thesis researches on web accessing sequence mining algorithm based on Markov chain. Through mining one web site log file, we can get users' behavior patterns, and predict their possible accessing pattern to this web site. This can help provide customized services to certain user group.On the basis of web site usability analyzing and log file mining, the thesis proposes a new algorithm based on Markov chain and modified PrefixSpan sequence pattern mining algorithm. Firstly, the thesis gives a deep analysis to the classical sequence pattern mining algorithm and its research status, and also discusses the problems when they are used to mine web sequence patterns. After introducing the property and application of Markov chain, the thesis proposes an analyzing method based on Markov chain. And then, a new sequence mining algorithm is proposed. The new algorithm constructs sequence database with one-step forward and backward transition probability matrix, and mines the database through combined bi-level and pseudo PrefixSpan projection. At last, the thesis uses an instance to analyze the algorithm performance and verify its effectiveness.Compared with classical sequence mining algorithm, the proposed algorithm can almost reach the accuracy with big advantages over efficiency. So it can provide better personalized services to certain kind of users in order to improve the web site usability.