Dynamic timeout-based a session identification algorithm

In order to improve the accuracy of data preprocessing in Web log mining, the basic procedure of data preprocessing is introduced firstly in this paper. Then the traditional session identification algorithm is fully analyzed, on the basis of which, a session identification algorithm based on dynamic timeout is presented. At the beginning of the algorithm, the initial timeout is computed for each page according to the statistical result, combining with the importance degree of page; then during the procedure of session identification, timeout is dynamically adjusted, and user sessions is determined judging by the dynamic timeout. Comparing experiment shows that the algorithm proposed can obtain a better performance on session identification.