Abstract
There are many different types of information to be considered when designing an information system, and a wide variety of modelling approaches and notations (or representations) have been developed to describe these different types of information. Some types of information are better expressed by some representations than others, so it is sensible to use multiple representations to describe a real-world phenomenon. Reconciling and integrating descriptions expressed using different representations is therefore an important part of the design process.
The objective of this research is to aid this reconciliation and integration within the context of information systems design. That is, to facilitate the use of multiple modelling representations for describing a phenomenon. To achieve this objective, the author has chosen an approach based upon translating descriptions of a phenomenon between different representations.
This thesis provides several important contributions in the area of information system design using multiple representations. Related work in the area is reviewed, and from this review is derived a terminology based on viewpoint-oriented methods that provides a consistent framework for the discussion of multiple representations.
Previous research into the use of multiple representations has focused on semantic data models. This is extended in this thesis to include diverse modelling representations such as functional dependencies and data flow modelling.
The process of translating between different representations is explored in depth, and several important issues identified. Translations are defined by a collection of rules that specify the mappings between constructs of representations. An abstract notation is developed for expressing these translations, and an extended version of Amor’s (1997) View Mapping Language is defined for the purpose of building detailed rule specifications.
Heuristics and enrichment are proposed as two methods of improving the quality of translations. A measure for determining the relative quality of translations is developed to show that heuristics do indeed enhance translation quality.
In addition, a method is developed for using translations to highlight potential design inconsistencies by translating descriptions expressed using different representations into the same form and comparing them.