Querying Uncertain Multiple Sources

摘要

Information often comes from multiple sources that may be conflicting. This makes uncertain the answers of a query to a set of sources. Possibility theory-based approaches to the handling of uncertainty in databases have been proposed and developed for a long time, in the case of a unique source of information. A multiple source counterpart of possibility theory has been recently proposed. Possibility and necessity set functions are then valued in terms of (possibly fuzzy) subsets of sources. Uncertainty may be assessed here either in terms of global reliability levels of sources or of tuples inside a source. When each source contains precise attribute values, each tuple is associated with the subset of sources that supports it as being an answer to a considered query, and with the subset of sources according to which the tuple is not an answer. In fact, these subsets of sources are fuzzy as they reflect the reliability levels. The benefit of the approach is to rank-order the answers to a query on a qualitative basis, in terms of subsets of sources and reliability levels.