A Novel Framework for Web Log Mining using Transductive SVM Classifier and Ontology based Associative Classification

Abstract

The web has revolutionized the concept of communication and interaction It offers new ways of business to business and business to customer transactions new mechanisms for newlineperson to person communication new means of discovering and obtaining information services and products electronically The rapid ecommerce growth has made both business community and customers face a new situation Due to intense competition on one hand and the customers option to choose from several alternatives electronic business community has realized the necessity of intelligent marketing strategies and relationship management newlineExploring user browsing characteristics using web usage mining techniques is a method that is used for this purpose This research works analyzes users browsing historical data stored in web log for next web page prediction The proposed system consists of three main steps, namely preprocessing potential user identification and next page prediction The preprocessing step takes as input the raw web log data and procedures a file that is more suitable for prediction This involves various steps like cleaning user identification and session identification Cleaning is the process of removing unwanted and irrelevant data User identification is the process of recognizing the unique users of the website This was newlineperformed using the IP address In this research work session identification using acyclic graph structures were analyzed Initially the browsing time of each page calculated using the timestamp attributes was estimated which were then discretized according to the length of browsing time These discretized values were then used to construct acyclic graphs Four types of acyclic graphs were analyzed They are Directed Acyclic Graph DAG Hierarchical newlineDirected Acyclic Graph HDAG Partial Ancestral Graph PAG and Mixed Ancestral Graph MAG Using graph mining algorithm DIGDAG algorithm the browsing sequences were extracted Phase II of the study focus on reducing the size of the web log data file by retaining only potential users For