Construction of Community Web Directories based on Web usage Data

TL;DR

This paper presents a method for constructing community-specific web directories using web usage data and clustering, demonstrating its effectiveness on real and artificial directories.

Contribution

It introduces a novel clustering approach and a combined criterion for building community web directories based on user access logs.

Findings

Effective community web directories were constructed from access logs.

The new clustering method outperforms existing approaches.

Results show high usability of the generated directories.

Abstract

This paper support the concept of a community Web directory, as a Web directory that is constructed according to the needs and interests of particular user communities. Furthermore, it presents the complete method for the construction of such directories by using web usage data. User community models take the form of thematic hierarchies and are constructed by employing clustering approach. We applied our methodology to the ODP directory and also to an artificial Web directory, which was generated by clustering Web pages that appear in the access log of an Internet Service Provider. For the discovery of the community models, we introduced a new criterion that combines a priori thematic informativeness of the Web directory categories with the level of interest observed in the usage data. In this context, we introduced and evaluated new clustering method. We have tested the methodology using access log files which are collected from the proxy servers of an Internet Service Provider and provided results that indicates the usability of the community Web directories. The proposed clustering methodology is evaluated both on a specialized artificial and a community Web directory, indicating its value to the user of the web.