A 5 Layer Information Expressivity Model Applied to Semantic Heterogeneity in a Decentralised Organisation

This paper discusses some issues associated with systems evolution in a decentralised organisation. The divisions of the particular organisation are functionally very similar but territorially decentralised and managerially autonomous. The paper considers the real case of an engineering fault logging system which has different versions in use by the different divisions of the organisation. It reviews the limitations of the standard architechtural approachs in the light of work on semantic heterogeneity. A recently proposed five layer model of information expressivity provides the theoretical framework for classifying the system variants. This approach allows us to focus on the layer requiring essential changes when seeking to interoperate these systems. We have chosen two examples from our case study; one example requires changes beginning at the semantic (business rule) layer and the other requires changes beginning at the schema (attribute representation) layer. The examples show the success of the five layer model in disentangling the semantics involved in this type of version change system interoperation.