FM-WAP Mining: In Search of Frequent Mutating Web Access Patterns from Historical Web Usage Data

Recently, a large amount of work has been done in web access pattern (WAP) mining. Most of the existing techniques focus on mining WAP that occur frequently from snapshot web usage data collection. However, web usage data is dynamic in real life. The dynamic nature of web usage data leads to two challenging problems. The first problem is the maintenance of existing WAP mining results. The second problem is discovering hidden knowledge from the historical changes to WAP. As the frequent WAPs are useful in many applications, knowledge hidden behind the historical changes to WAPs, which reflects how WAPs change, is also critical to some applications such as adaptive web, web site maintenance, business intelligence, etc. In this paper, we propose a novel approach to discover hidden knowledge from historical changes to WAPs. We define a new type of knowledge, Frequent Mutating WAP (FM-WAP), based on the historical changes to WAPs. Moreover, a FM-WAP mining approach, which can discover all the FM-WAPs efficiently from historical web usage data, is presented. Firstly, the historical web usage data collection is partitioned into different groups according to the user-defined calendar pattern, where each group is represented as a WAP forest. Then, changes among the WAP forests sequence are detected and stored in the global forest. Lastly, the FM-WAP is extracted by traversing the global forest. Extensive experimental results show that our proposed approach is efficient with good scalability to very large web usage data collection. Moreover, this approach can produce new knowledge about web access patterns that cannot be discovered by existing techniques.