Abstract:

The web 2.0 is now one of the best access point to any type of information. Actually, its online sources, since years, are an extraordinarily rich and important base of knowledge and information. Of course, this richness and diversity in terms of quantity and quality has dangers and abuses that must be identified and avoided.

As part of this research, the wiki system is a powerful tool for those who want to share their knowledge, collaborate and learn in the purpose of having a relevant background of information acquired during participation.

The web 2.0 is characterized by flexibility and interactivity, especially at the interfaces, such as the wiki interface, that allow users to interact and express their opinions. Nevertheless, each system has its limits. Consequently, the research on the contributions performance of individual users is still unexplored because it is partly related to the design of wikis, which is designed for collaborative work. This has made the task of evaluation and assessment of individual contributions difficult.

In this study, we will highlight the importance of identifying the relevant articles based on the opinions of users and their contributions.

In this way, we will try to extract informations from these users interactions to assess the relevance of articles published online, using the data mining techniques to form groups of articles in an automatic way.The area of application is wiki content and it is different from the usual data mining studies. In fact, the use of the data mining technique in wiki field demonstrates more varied and significant findings, and may lead us to reach our goal : showing the quality of the content wiki.

This study utilizes data mining to discover the type of wiki articles. Cluster analysis and K-means analysis are used as data mining techniques. Furthermore, the experimentation shows the advantages of using discretization techniques to get results with a value of error equal to zero.

The data mining process is carried out and explained in detail, and consists of creating a specific number of groups, depending on our needs. So, the selection of attributes has a very important role to play in improving the result of clustering.

Keywords:

Web 2.0, Wiki, Data mining, Cluster Analysis, K-Means Algorithm, content reliability, Discretization, weka.