Selecting Discriminative Facets for Guided Navigation
External - Copyright Consideration
Keyword(s): information retrieval; adaptive interfaces; faceted browsing
Abstract: Faceted metadata browsing systems offer a useful guided navigation interface to information seekers to specify the parameters of data items that they seek. For the most part, current faceted search/browse interfaces are manually created and static and display all facets. This can make them harder to use while navigating unfamiliar, dynamically changing, and/or heteregenous data collections, where not all items share the same metadata attributes. They can also be somewhat complex, especially for rendering on restricted devices such as mobile handheld clients. In this paper we argue for automatic, dynamic methods to select the best facets for a given collection and to present these as the only, or as the preferred, facets to the user. We describe a method for facet selection using discriminative score computation, including clustering metrics, to evaluate facets. An implementation using digital image metadata is described.
Back to Index