Semantic model driven engineering

Thesis

Semantic interoperability is a concern in the design and implementation of modern information systems. Whenever two systems communicate it is important that they share an adequate, common interpretation of any information transferred. Similarly, whenever data created or ac- quired in one context is used in another, it is important that the provenance of the data is consistent with its intended usage.

Medical researchers depend on understanding good quality data, most of which is kept in information processing systems designed for multiple purposes by multiple organisations and individuals with different motivations in mind. As a result the original meaning and context of the data being used in analysis can be difficult to establish, in addition there are many ways in which data can be corrupted, missing or otherwise altered, resulting in research data scientists spending the majority of their time in activities variously referred to as ‘data-cleaning’, ‘data- wrangling’ and ‘data-munging’, sometimes leading to a lack of trust in the quality of the data.

This thesis is a response to that challenge. The approach combines and builds upon existing work in two distinct areas. The first is that of data standards and metadata registries, in which a consensus has been reached regarding the abstract representation of the ‘semantics’ of a data element. The second is that of software modelling and model-driven engineering, in which a consensus has also been reached regarding the abstract representation of data structures and the relationships between data elements.

The key contributions of this thesis are firstly a formal language for reasoning about se- mantic interoperability in terms of data models. In addition a formal grammar is provided on which metadata registries can be built guaranteeing consistency of models. A modelling language for metadata description is introduced, based upon the widely-used Eclipse modelling framework, a de-facto standard in model-driven engineering. The thesis then demonstrates that this language, used in combination with a suitable set of constraints for metadata registration and management, can be used to achieve and maintain semantic interoperability in practice.

Authors: Milward, D

• DOI: 10.5287/ora-nzrmw5j44
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: Software Modelling; Model Driven Engineering; Formal Methods
• Subjects: Formal methods (Computer science)--Congresses; Computer Science