Improvement of the Scalability for Web Log Mining with LCS

摘要

Nowadays, information distribution via websites is one of the most important issues. Therefore, website administrators are required to understand access trends of the websites properly. We investigate a method proposed in our previous work for mining access logs with LCS (Longest Common Subsequences) to extract frequent access sequences. However, the method still has a problem of the scalability. The execution time increases for large websites. In this paper, we propose an approach of filtering sequences with a hash function to improve the performance for analyzing large websites. We evaluate the efficiency of the approach using access logs of a real website and artificial data.