A theoretical and practical investigation of tools and techniques for the structuring of data and for modelling its behaviour
This thesis is about data and behaviour modelling for information system development. It has been sponsored at different times by two specialist consultancies: CACI Inc International and James Martin Associates. Initially I found problem areas in the field of system development by interviewing practitioners and by consultancy. These initial problem areas were whittled down to: action modelling, entity model clustering and a diagrammer. Action modelling is the modelling of detailed data behaviour using the same structuring concepts as data modelling. It was developed because of a lack of such analysis in systems development. Entity model clustering is about aggregating the entity types in a large entity model to abstract the essential meaning and to identify the most fundamental entity types. It was developed because of a need to summarise large entity relationship models for usability and comprehension. It has been used widely and has many benefits. A parallelism between data and activity modelling was developed as a result of the research into action modelling and entity model clustering. It needed the concepts derived from the other two areas to finally complete the theory, summarised as: every data modelling concept and structure has an exact equivalent in activity modelling and vice-versa. This theory gives a wholeness and completeness to modelling data and activity. A diagrammer was produced for the automatic production and manipulation of entity relationship diagrams from a base description. These diagrams are the basic tool of the data modeller; automating them saves time and potentially raises their accuracy. The main research problem was that few companies were willing to be guinea pigs, so most of the research was developed by thought 'games'. Most areas have been published in refereed publications as this was seen as the best way of establishing their academic credibility. All areas have been incorporated into or had an impact on James Martin Associates and their methodology Information Engineering, which provides a framework for coordinating the research areas. This research can best be techniques for improving summarised as the systems an attempt to find analysis process.