Mining System Based on a Hybrid Approach of Techniques JOSE AGUILAR, YENNY GUERRERO CEMISID , Facultad de Ingenieria Universidad de los Andes Nucleo La Hechicera, Merida VENEZUELA Abstract: - The Web Mining arises like an appropriate tool to exploit the derived knowledge of the web-user interaction, describing models that use patterns and characterize the profiles of the different groups of users which use Internet. To achieve this, currently there are numerous techniques. Some of these techniques are integrated in this work to build a Hybrid System of Web Mining that allows to extract useful information of the web users. In this way, we try to exploit the capabilities of each one. Specifically, three techniques of the area of Web Mining were used: Sequential Patterns, Path Analysis and Cubes. The System obtains a group of access patterns from the users to a website, to arrange them in a multidimensional structure, called Cube. Using that, the system can discover correlations between the web pages and users' groups, behaviors of the web users, among other things. Key-Words: - Web Mining, Web Computing. 1 Introduction The explosive growth of Internet has made more necessary for the users to use automatic tools to find, to extract, to filter and to evaluate the available resources of information over it. There is an infinity number of places that we need to visit and classify when making a search. There are powerful search tools to find information for category or for content, such as Altavista, Yahoo, Google, etc. For these searches we need to introduce keywords, and they determine the web pages that contain these words, trying to satisfy the user's requirements. Many times, these consultations bring inconsistence, or documents that fulfill the search approach but not with the users interest [6]. There is necessity of having new technologies that help us in our search processes and, even more, of technologies that help us to use the content of the web more efficiently [5,6]. For this reason, in the last years a series of techniques that allow the advanced processing of data on the Internet have been developed. These techniques carry out a depth analysis in an automatic way, and they belong to an area denominated Web Mining [1, 2, 6]. The web mining is an area that involves the use of techniques based on the data mining, guided for the discovery and automatic extraction of information, of documents and services in the web. It should be kept in mind that the human beings play an important part in the process of knowledge or information discovery in the web, since the web is an interactive media. The web mining provides tools so that the user can discover and exploit the implicit knowledge in the web. The main axis of the approach proposed in this work involves the use of techniques of the area of Web Mining like: Cubes, Sequential Patterns, Path Analysis, in order to propose a System of Mining Web that is a hybrid of them, in such a way of exploiting the advantages of each one. Our Hybrid System of Web Mining can be used to extract the useful information for the web users, such as [6]: correlate between the web pages and users' groups, behavior of the users when navigating for Internet, cluster of pages according to the users, among other things. 2. Web Mining The Web Mining can be defined as the automatic analysis and discovery of useful information from documents and services of the web [5,6]. The Web Mining is an area that involves the use of techniques based on Data Mining, but now guided to the discovery and analysis of the information in the web. The techniques of Web Mining can be used to access more efficient to the information contained in the web, of direct or indirect form. Inside this definition three approaches exist [6, 7]: Web Content Mining: it refers to the automatic search of information and extraction of knowledge, based on the content and the descriptions of the documents in the web. Web Structure Mining: it refers to the process of inferring knowledge, based on the organization and the references or connections among documents of the web. Proceedings of the 10th WSEAS International Conference on COMPUTERS, Vouliagmeni, Athens, Greece, July 13-15, 2006 (pp206-211)