The problem of integrating information from con icting sources comes up in many current applications, such as cooperative information systems, heterogeneous databases, and multi-agent systems. We model this by the operation of merging rst-order theories. We propose a formal semantics for this operation and show that it has desirable properties, including abiding by majority rule in case of con ict and syntax independence. We apply our semantics to the special case when the theories to be merged represent relational databases under integrity constraints. We then present a way of merging databases that have di erent or con icting schemas caused by problems such as synonyms, homonyms or type con icts mentioned in the schema integration literature.
Jinxin Lin, Alberto O. Mendelzon