Applying Semantic Web Usage Mining in Prediction System

Abstract

Due to uncontrolled exponential growth in web data, knowledge based retrieval has become a challenging task. The one viable solution to the problem is the merging of conventional web mining with semantic technologies. There are many methods based on web usage mining for prediction system but most of all are based on traditional methods such as sequential pattern mining, clustering and so on. In this research, the reference ontology is built according to the web site structure and the domain ontology is also built according to the clean web log files. Then the domain information is generated and ontology based Perfect Hashing and Shrinking (PHS) algorithm is used in developing sequential pattern. Moreover, the prediction list is produced by applying semantic similarity related with the domain ontology.