Mining uncertain web log sequences with access history probabilities

This paper proposes (1) modeling uncertainty in web log sequences using the most recent periodic web log which attaches computed existential probabilities between 0 and 1, to events in the sequences, (2) using the newly proposed uncertain PLWAP web sequential miner for these uncertain access sequences. While PLWAP only considers a session of web logs, U-PLWAP takes more sessions of web logs from which existential probabilities are generated and there is the need to traverse each suffix tree from the root in order to scan for existential probabilities of items already found along the path. Experiments show that U-PLWAP is faster than U-Apriori, and UF-growth. © 2011 Authors.