Akademik Veri Yönetim Sistemi
8th International Conference on Hybrid Artificial Intelligent Systems (HAIS), Salamanca, Meksika, 11 - 13 Eylül 2013, cilt.8073, ss.181-190
Sequential pattern extraction is essential in many applications like bioinformatics and consumer behavior analysis. Various frequent sequential pattern mining algorithms have been developed that mine the set of frequent subsequences satisfying a minimum support constraint in a transaction database. In this paper, a hybrid framework to sequential pattern mining problem is proposed which combines clustering together with a novel pattern extraction algorithm that is based on an evaluation function, which utilizes user-defined criteria to select patterns. The proposed solution is applied on Web log data and Web domain, however, it can work in any other domain that involves sequential data as well. Through experimental evaluation on two different datasets, we show that the proposed framework can achieve valuable results for extracting patterns under user defined selection criteria.