Architecture of an hybrid system for experimentation on Web information retrieval incorporating clustering techniques

The results retrieved from common search engines are frequently not acceptable because of the large amounts of Web pages that are returned and the ambiguity of the results. Several authors have suggested grouping the Web pages by topics using clustering techniques. The application of these techniques involves different alternatives in different areas. Carrying out these research works could be easier and their cost lower if an experimentation hybrid system could be available that implemented the different alternatives or new ones could be added with a low cost. This work proposes an architecture of a Web information retrieval experimentation hybrid system that incorporates clustering techniques. This hybrid system is composed of different replaceable components, so that we could test different alternatives with a low cost. Later, using the proposed and the implemented system an empirical study is exposed in order to evaluate what type of clustering techniques is the most appropriate.