Document Clustering for Mediated Information Access

This paper addresses the problem of accessing very large heterogeneous document collections by proposing a new approach to using clustering for information retrieval: mediated access through a clustered collection . In what is actually an information access environment, the user can explore a relatively small, well structured, pre-clustered collection covering a particular subject domain, in order to understand the concepts encompassed and to clarify and refine his/her information need. The user can ostensively indicate clusters and documents of interest and be assisted in formulating a query, based on which a search can be done on a large, non-structured collection. Finally, the original cluster structure is the basis for visualisation tools that allow the user to explore search results. WebCluster, the system implementing these ideas, is presented, together with results of an initial formative experiment and plans for future experiments.