| With the proliferation of websites and increased E-commerce systems, maintaining web log data has become very important for multitude of services. Web log data provides significant information about users behaviours which are used by companies to improve their websites and customized services. The insights from web log mining can help improve website design and personalization strategies. In web usage mining, about 80% of the time is spent on data preprocessing. Parallel computing techniques can improve the performance and reduce the time taken for web usage mining. In this thesis, we propose to use parallel computing techniques to preprocess raw web log data and apply parallel association rule mining algorithms to extract user access-patterns from web log data. The results of the implementation are promising for future work in web usage mining. |
