The Importance of Document Ranking and User-Generated Content for Faceted Search and Book Suggestions

In this paper we describe our participation in INEX 2011 in the Books and Social Search Track and the Data Centric Track. For the Books and Social Search Track we focus on the impact of different document representations of book metadata for book search, using either professional metadata, user-generated content or both. We evaluate the retrieval results against ground truths derived from the recommendations in the LibraryThing discussion groups and from relevance judgements obtained from Amazon Mechanical Turk. Our findings show that standard retrieval models perform better on user-generated metadata than on professional metadata. For the Data Centric Track we focus on the selection of a restricted set of facets and facet values that would optimally guide the user toward relevant information in the Internet Movie Database (IMDb). We explore different methods for effective result summarisation by means of weighted aggregation. These weighted aggregations are used to achieve maximal coverage of search results, while at the same time penalising overlap between sets of documents that are summarised by different facet values. We found that weighted result aggregation combined with redundancy avoidance results in a compact summary of available relevant information.