Improving interoperability by using a three data model approach

Achieving interoperability in multidatabase systems is unlikely to be achieved without a full understanding of component database systems' semantics. The full semantics of many database applications cannot be understood without combining the database's data semantics from a logical model of data L with data semantics embedded in a file-system model F and in a physical storage model P. To prove this assertion, the authors use proof by contradiction on its negation. Two examples are provided to demonstrate false positives, multidatabases with identical logical models L which are not interoperable, one due to differences in F and the other in P. One example is provided to demonstrate a missing value, an interoperable database without an explicit logical model L, but rich in semantics desirable from its file-system model F. A technique is proposed for describing how semantics embedded in L, F and P can be combined.<>