Behaviour Recovery and Complicated Pattern Definition in Web Usage Mining

Data mining includes four steps: data preparation, pattern mining, and pattern analysis and pattern application. But in web environment, the user activities become much more complex because of the complex web structure. So user behaviours recovery and pattern definition play more important roles in web mining than other applications. In this paper, we gave a new view on behaviour recovery and complicated pattern definition. We used several methods to recover different user behaviours, such as simple behaviour, sequence visiting, tree structure behaviour, acyclic routing behaviour and cyclic routing behaviour. Based on various recovered behaviours, we raised how to define complicated usage patterns. These usage patterns include constraint association rules, constraint sequential patterns, deepest access paths, shortest access paths, tree structure accessing patterns, parallel visiting patterns, circle visiting patterns and so on. We also gave some experiment results about these complicated access patterns which reveal some interesting usage behaviours.