A formal, scalable approach to semantic interoperability

Scientific progress is increasingly dependent upon the acquisition, processing, and analysis of large volumes of data. The validity of results and the safety of applications rely upon an adequate understanding of the real-world semantics of this data: its intended interpretation, and the context in which it is acquired and processed. This presents a challenge: interpretations vary, context is infinite, and either may change over time. This paper addresses that challenge. It introduces a language for the description of real-world semantics that allows for multiple, evolving interpretations, together with a high degree of automation in the capture and creation of contextual metadata. The language itself has a mathematical semantics, and supports a notion of semantic interoperability closely related to existing, formal notions of refinement. The language represents a scalable approach in three respects: it is compositional, in terms of composing real-world semantics piece by piece; it allows for multiple perspectives, allowing the parallel development of different interpretations; and it supports automatic transformations to and from implementation languages. The practical application of the approach is illustrated with examples from large-scale medical research.