Uncertainty in Data Integration

Abstract

Data integration has been an important area of research for several years. In this chapter, we argue that supporting modern data integration applications requires systems to handle uncertainty at every step of integration. We provide a formal framework for data integration systems with uncertainty. We define probabilistic schema mappings and probabilistic mediated schemas, show how they can be constructed automatically for a set of data sources, and provide techniques for query answering. The foundations laid out in this chapter enable bootstrapping a pay-as-you-go integration system completely automatically.