Abstract: Web mining is application of data mining which is useful to extract the knowledge. So, wecan use web mining algorithm in understanding users’ navigation behavior. In the proposed architecture first step is the preprocessing of web log data. In pre-processing step, modified data cleaning algorithm removes all irrelevant entries from weblog file. After data cleaning step user identification is performed depends on user's domain name or IP addresses. After preprocessing step, web mining algorithms of proposed architecture are applied to study the user navigation behavior. In the proposed architecture three algorithms are proposed. Those algorithms are, Modified Data Cleaning Algorithm, Proposed Modified Clustering Algorithm-I and Proposed Algorithm-II based on preferred path mining. Modified data cleaning algorithm is uses to remove all irrelevant entries and all multimedia files from weblog file. Proposed modified clustering algorithm-I is adapts to cluster users based on their similarity. Proposed algorithm-II based on preferred path mining accomplishes path mining on the clusters of different users found in clustering step. The results of proposed algorithms show improvement in memory usage and execution time.
Keywords: Web Usage Mining, Pre-Processing, Weblog File.