An Improved Web Log Mining and Online Navigational Pattern Prediction

The aim of this study is to improve web log mining and online navigation pattern prediction. Web mining is an active and wide area which incorporates several usages for the web site design, providing personalization server and other business making decisions etc. Efficient web log mining results and online navigational pattern prediction is a tough process due to vast development in web. It includes the process such as data cleaning, session identification and clustering of web logs generally. In this study initially the web log data is preprocessed and sessions are identified using refined time-out based heuristic for session identification. Then for pattern discovery a density based clustering algorithm is used. Finally for online navigation pattern prediction a new technique of SVM classification is used, which rectifies time complexity with increased prediction accuracy.