Enhanced DBSCAN with Hierarchical Tree for Web Rule Mining


Neelima Gullipalli
Sireesha Rodda


Like other mining, web mining is also necessary to increase the power of web search engine to identify the intended web page and web document. While processing with large datasets, there arises several issues associated with space availability, similarity relationships between different webpage’s and running time. Hence, this paper intends to develop an enhanced web mining model based on two contributions. At first, the hierarchical tree is framed, which produces different categories of the searching queries (different web pages). Next, to hierarchical tree model, enhanced Density-Based Spatial Clustering of Applications with Noise (DBSCAN) technique model is developed by modifying the traditional DBSCAN. This technique results in proper session identification from raw data. Moreover, this technique offers the optimal level of clusters necessitated for hierarchical clustering. After hierarchical clustering, the rule mining is adopted. The traditional rule mining technique is generally based on the frequency; however, this paper intends to enhance the traditional rule mining based on utility factor as the second contribution. Hence the proposed model for web rule mining is termed as Enhanced DBSCAN-based Hierarchical Tree (EDBHT). It benefits in providing the search results depending on high level information (e.g., location), so that the ability of search engine in providing the interesting association rules can be improved. Next, to the implementation, the performance of proposed EDBHT is found to be enhanced when compared over several traditional models.


Special Issue Papers