Jeudi 19 février de 14h00 à 15h30
Auditorium IRCICA, parc scientifique de la Haute Borne à Villeneuve d’Ascq
Several distinct research areas related to data management and the web deal with answering an information need by making use of semantic information about the data. Examples include the Semantic Web, data integration, and semantics-based optimization modules for commercial database systems. This talk describes work on creating a unified platform for answering queries that subsumes all of these problems. The vision is to support distributed datasources connected by integrity constraints, supporting a wide variety of access models, cost models, data models, constraint languages, and query languages. We will describe a step towards this goal, embodied in the PDQ (Proof-driven querying) system at Oxford. PDQ supports a rich language for constraints and queries, along with the ability to describe several kinds of interfaces for data access. It extends traditional cost-based query optimization to the setting of constraints, and provides an architecture that allows one to « plug in » different reasoning systems to answer queries. The talk will give a quick look at the system in action, a bit of the underlying theory. It also discuss where we stand in trying to fit together reasoning about data, cost-based query optimization, and modeling of interfaces and access.
This includes work with Balder ten Cate, Julien Leblay, Efi Tsamoura, and Michael Vanden Boom.