A compact representation for efficient uncertain-information integration

Abstract

The probabilistic relation model has been used for the compact representation of uncertain data in relational databases. In this paper we present the extended probabilistic relation model, a compact representation for uncertain information that admits efficient information integration. We present an algorithm for data integration using this model and prove its correctness. We also explore the complexity of query evaluation under the probabilistic and extended probabilistic models. Finally, we study the problem of obtaining a (pure) probabilistic relation that is equivalent to a given extended probabilistic relation, and present approaches and algorithms for this task. This work is the first and critical step towards practical and efficient uncertain information integration.