Classification Rules for Pre-Analysis Filtering of Web Transactions
Omar H. Karam, Ahmed Mahmoud Hamad, W.H. Riad · IEEE International Conference on Computer Systems and Applications, 2006. · 2006
Web usage mining is the process of applying data mining techniques to discover usage patterns from web data to serve the needs of web based organizations. These patterns are usually expressed in the form of association rules. In this paper, we attempt to improve the performance of algorithms for mining web association patterns in terms of runtime and quality of results. We made use of the fact that only a small portion of user sessions lead to the generation of interesting rules. Decision trees were used to extract classification rules describing interesting sessions. These rules were applied on future data to focus the analysis on interesting sessions. The average classification accuracy was 94.29% and the average sensitivity was 28.16%. The use of classification rules caused an average improvement in runtime of 63.13%. The improvement in the precision of the system was not significant. The average improvement in precision was 5.34%. The average recall of the system was 95.06%.