A Novel Weighted Support Method for Access Pattern Mining

Sequential Pattern Mining is an important data mining technique that finds out all frequent sequential patterns in a sequence database. Applications in wide range of important domains make Sequential Pattern Mining an interesting area of research. Conventional approach for sequential pattern mining treats each and every item in the sequence with equal importance and thus fails to reflect the individual significance of items. Weighted Sequential Pattern Mining is an approach that treats different items in the sequences with different weights so as to reflect the importance of each item. Thus, weighted method models real life sequence database in a better manner and more efficient than the conventional sequential pattern mining. Weighted sequential pattern mining can be used to mine web access patterns more efficiently from web log data. This paper proposes a new weighted access pattern mining algorithm to mine weighted access patterns in a web log database. The proposed method uses frequency of user visit to give weights to web pages during the mining process. Through extensive experimental evaluation the algorithm is proved to be promising.