Fast Mining of Maximal Web Navigation Patterns

Discovering user navigation patterns in web log sessions has been an interesting problem and used with many applications including web site development, e-business, e-learning etc. Most of the proposed algorithms for mining web log patterns generate candidate sequences and test whether they are frequent or not, based on the given min-sup. In this paper, we present a fast method that aims at mining prefix based maximal contiguous sequence patterns without generating candidate sequences level-by-level. It first generates maximal potential sequences and mines only them in the database using minimized search space.   Performance evaluation of the proposed algorithm is done by conducting experimental studies on a real dataset and found satisfactory when compared to previous approach.