Value-Based and Representation-Based Querying of Possibilistic Databases
Patrick Bosc, Laurence Duval, Olivier Pivert · Studies in fuzziness and soft computing · 2000
In this paper, we address the issue of querying imperfect data represented by possibility distributions. We distinguish between two types of queries: those involving conditions on the values, and those involving criteria on the representations of ill-known values. These two approaches are successively considered. We first recall some classical results relating to the querying of databases involving null values, and we point out the problems that arise in the specific context of disjunctive weighted data when value-based queries are dealt with. The necessity of defining a typology of relevant queries is emphasized. Then, we introduce a new querying framework allowing to handle ill-known data at a representation level. This framework, based on the notion of a weighted set, offers an alternative solution to the use of value-based queries and thus could be used to extend the querying capabilities of database systems aimed at handling ill-known values.