Degree Name

Doctor of Philosophy


School of Computer Science and Software Engineering


There exist many approaches to agent-based conflict resolution. Of particular interest is on approaches that adopt argumentation as their underlying conflict resolution machinery. One such approach is to view the argumentation process as an approach to attack, persuade an opponent or to defend one’s belief. This dissertation proposes an abstract accrual argumentation framework that re-evaluates the treatment and utilisation of preference values within argumentation. We firstly present an incremental improvement on existing work to capture accrual of arguments within an abstract argumentation framework. Drawing from the incremental improvement, this dissertation highlights the importance of information source in argumentation and the effect on agent’s decision making during argumentation. In most argumentation systems, the argument source plays a minimal role. We feel that ignoring this important attribute of human argumentation process reduces the capabilities of current argumentation systems. Secondly, this dissertation identifies the need for justification management in a setting where multi-agent performs negotiation or argumentation. An outcomedriven justification management framework is proposed in which traditional approaches in argumentation are modified to assist in the elicitation and management of justifications hence permitting the novel conception of mixed-initiative argumentation. Finally, the framework is also evaluated in the context of a clinical group decision support in medicine.



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.