EN FR
EN FR


Section: Research Program

Representative Semantic Query Answering

Top-k search is a classical topic, studied in relational databases, semantic web, recommandation systems,... It is extremely useful, among other, when a human user face a large number of query results, allowing the user to reformulate the query if necessary. However, we argue that top-k search incurs a bias on the perception of the set of results which is out of the control of the user. Our goal is to provide the user with k answers as well which are chosen so as to represent the diversity of the answer set. We will first consider this problem in the setting of relational or RDF databases. We will then extend to more heterogeneous sources, including in particular plain text.