Proposed Clustering Model Based On Sequential Rules In The Web Mining

Abstract

The important growth of the pages contained in the website as well as the number of users navigating requires research tools that allow studying the behaviour of users in the Web Usage Mining. Among these tools, clusters analysis is considered as the most important technique in this area. Based on this technique, several methods have been developed; the most popular is the partitioning method. However, its principle, as it appears to be unsuitable for web data, represents a sequential data stream where the similarity notion must be taken into account when calculating the distance between objects.

This paper attempts to overcome the limitations of this method and proposes a new user clustering model. The proposed approach is based on the extraction of sequential patterns along with the generated sequential rules. The experimental realization has been carried out by implementing the proposed algorithm and the k-medoids partitioning algorithm. This study is carried out with the aim of comparing the performance relative for each of them through a set of measures that help evaluate the quality of the generated clusters.