Understanding Users' Subject Interests in the Web Site Based on Their Usage of Its Content: A Novel Two-Phase Clustering Framework

摘要

In order to understand the behavior of website users, a deep analysis of content and usage data can reveal valuable knowledge about the main subjects these visitors are truly interested in. Preprocessing and clustering the highly unstructured content of web pages should be addressed very carefully in order to provide effective results. In this paper, a novel proposed two-phase self organizing feature map clustering framework to segment web users based on their subject interests in the diverse content of a University website is described. Also, the overall noise and dimensionality reduction of the sample web site content is properly addressed through the formulation of a comprehensive ten-step preprocessing procedure, which provided very promising experimental results when applied to the input web pages in the first phase of the proposed framework.