Auto-Suggest Function Integrated with Faceted Search: User Evaluation and Application to Real-life Systems

Abstract

Faceted search is becoming popular as a dynamic navigation method that allows users to interactively explore and search complex information spaces. Since the faceted search, which follows predefined classification categories, is not necessarily easy for users to express diversifying search needs, a dynamic term suggestion or auto-suggest interface would be of great help to formulate better search queries. We have proposed a term-weighting method that does not rely on existing domain vocabulary in our previous stuty. In this study, we conducted experiments to evaluate the proposed method under the practical condition where faceted search and dynamic term suggestion can be combined in any order. In particular, the experiments were made by using two faceted-search systems available in public for multilingual school documents and course materials. As results of the experiments, it was concluded that the proposed method was effective in terms of search precision for the initial query as compared with a previous method that relies on a particular domain vocabulary at hand. Although the auto-suggest method by itself had limitation in the search success rate, higher success rate was achieved when the users were provided with faceted search as well as auto-suggest function. Finally, on the basis of the results, an auto-suggest function with the proposed weighting method was integrated into the real-life systems, and was opened to the public.