Integrating and managing conflicting data

Abstract

Data integration aims to provide a uniform integrated access to multiple heterogeneous information sources, designed independently and having strictly related contents. The heterogeneity among sources may range from the hardware and software platforms to the data model and the schema used to represent information. In this paper we consider data inconsistencies and deal with the integration of possibly conflicting instances coming from different sources and related to the same concept. In order to support the data integration process, we introduce some integration operators for combining data coming by different sources, preserving the information provided by each source separately. Generally, a view obtained by integrating heterogeneous sources could contain instances which are inconsistent, w.r.t. some integrity constraints defined on the integrated view schema. Therefore, we present a technique which allows us to compute consistent answers to queries involving possibly inconsistent data. © Springer-Verlag Berlin Heidelberg 2001.