Doctor of Philosophy
School of Information Technology and Computer Science
Krishna, A., Practical methodologies for agent-oriented conceptual modelling, Doctor of Philosophy thesis, School of Information Technology and Computer Science, University of Wollongong, 2006. http://ro.uow.edu.au/theses/519
Many modelling techniques tend to address “late-phase” requirements while many critical modelling decisions (such as determining the main goals of the system, how the stakeholders depend on each other, and what alternatives exist) are taken during early-phase requirements engineering. The i* modelling framework is a semiformal agent-oriented conceptual modelling language that is well-suited for answering these questions. This dissertation addresses two key challenges faced in the practical deployment of agent-oriented conceptual modelling frameworks such as i*. First, very little exists by way of principled methodologies for eliciting agent-oriented conceptual models. The high-level and abstract nature of these models poses special challenges in devising elicitation techniques. Second, there has been limited adoption of agentoriented conceptual modelling techniques in industry. We address the first challenge by developing a novel elicitation technique based on requirements capture templates. These templates can themselves be derived from enterprise/ domain ontologies and organisational models, and interact with these in interesting ways. Our approach to addressing the second problem is based on the observation that the value of conceptual modelling in the i* framework lies in its use as a notation complementary to existing requirements modelling and specification languages, i.e., the expressive power of i* complements rather than supplants that of existing notations. The use of i* in this fashion requires that we define methodologies that support the co-evolution of i* models with more traditional specifications. This research uses the notion of co-evolution in a very specific sense to describe a class of methodologies that permit i* modelling to proceed independently of specification in a distinct notation, while maintaining some modicum of loose coupling via consistency constraints. This research examines how this might be done with formal specification notations (specifically Z) as well as an industry standard modelling language (UML). Our aim, then, is to support the modelling of organisational contexts, intentions and rationale in i*, while traditional specifications of functionality and design proceeds in either the formal notation or UML. Much of this research has been motivated by issues arising from an industry-scale enterprise modelling exercise conducted for an emergency services agency. This research has been validated through a detailed case study involving a major government emergency services agency (who provided funding for this project). A very large-scale and comprehensive agent-oriented conceptual model of this agency and its requirements has been developed, and forms the basis for the validation component of this dissertation.