This paper mainly discussed Web log mining,the application of date mining to lot data generated by Web servers, which could assist the Webmaster to optimize site architecture and increase visiting efficiency. Based on the analysis of difficulties, the corresponding solutions of Web log mining and the architecture of SWLMS, our sample Web log mining system was addressed. The data-preprocessing phase in SWLMS, including data cleaning, user recognition, session identification and path filling, was discussed in detail. Then, the sequential pattern recognition phase and its algorithms, including the recognition of maximum forward paths and frequent traversal paths, were presented.