A hybrid system for concept-based web usage mining

Article type: Research Article

Authors: Ríos, Sebastián A.^{a; *} | Velásquez, Juan D.^b | Yasuda, Hiroshi^a | Aoki, Terumasa^a

Affiliations: [a] Applied Information Engineering Laboratory, University of Tokyo, 4-6-1 Komaba Meguroku, Japan | [b] Department of Industrial Engineering, University of Chile, República 701, Santiago, Chile

Correspondence: [*] Corresponding author. E-mail: [email protected]

Abstract: A web site should be easy to browse by visitors. However, sometimes the reality is quite different. Situations like several unrelated topics in a single web page may lead to confusion and make harder to reach the information that the visitors are looking for. The design of the whole site (interface, content, structure, usability, etc.) is one of the most important aspects for any institution that wants to survive in the cyberspace. This work aims to extract new knowledge about how the visitors are using the web site and, in this way, to improve the site structure and content. We propose a hybrid system that combines the web usage mining (WUM) approach for analyzing the visitor browsing behavior and the conceptual classification of web pages for understanding the visitor text preferences in the site. The proposed system was tested in a real web site, which shows its effectiveness.

DOI: 10.3233/HIS-2006-3404

Journal: International Journal of Hybrid Intelligent Systems, vol. 3, no. 4, pp. 219-235, 2006