Interoperability is a requirement in today’s information systems, where differences among several component operating systems, database management systems, or pieces of hardware must become transparent to the end user. This paper presents a theory concerning semantic interoperability, whose goal is to ensure the proper exchange of data among heterogeneous information systems.
The basic concept of this theory is context information, which includes properties related to the usual data (for instance, the context of a datum representing a price may include its currency, a scale factor). Semantic value, which originated in LISP’s idea of a property list, is defined as the association between a context and a simple value. A data value, when exchanged from one system to another, migrates from its source context to a receiver context.
After an introduction, the second section provides the foundation for the theory of semantic values. Conversion functions, which realize the transformation of semantic values from one context to another, are presented, and characteristics useful in semantic comparison and arithmetic (such as total, lossless, lossy, and order-preserving conversion functions) are discussed. The authors examine a specific situation derived from semantic comparison, namely the “resolvable” and “unresolvable” properties.
In Section 3, the authors propose a system architecture that provides a general solution to semantic interoperability. Its central component, responsible for the exchange of values between several information subsystems, is the context mediator. Sections 4 through 6 apply this architecture to the relational model. A relational source database, a relational receiver, and the implementation of a context mediator are described. A specific manipulation language, Context-SQL, is introduced as an extension of SQL by which the context information of a semantic value can be manipulated.
The material is accurate and clear. The authors compare their theory with other issues in semantic interoperability, highlighting the way it complements previous work. Adequate references are provided.
The paper is intended primarily for researchers in semantic interoperability. It suggests several directions for future research; some are stated explicitly at the end of the paper, while others can be inferred from the exposition (the theory only applies to resolvable properties). The work will also be useful to companies implementing interoperability.