The integration of relationship instances from heterogeneous databases

In the process of integrating legacy databases, one has to resolve inter-database conflicts at both the schema and instance levels. In this paper, we discuss relationship conflicts as a special type of conflicts to be resolved during the database integration. Relationships are properties that relate real world objects. So far, most inter-database relationship conflicts are addressed at the schema-level by various schema integration techniques. However, instance-level relationship conflicts are largely neglected. This paper therefore investigates the causes of instance-level relationship conflicts and proposes a taxonomy for classifying instance-level relationship conflicts. In addition, we develop a systematic process to resolve instance-level relationship conflicts and incorporate the resolution steps into the overall database integration process. Instance-level relationship conflict detection algorithms have also been developed to aid the resolution process. This research should facilitate database integration work for both multidatabase and data warehousing approaches. Most importantly, it should improve the data quality of the integrated databases. (C) 2000 Elsevier Science B.V. All rights reserved.