How do you make software components argue? An exploration of argumentative reasoning within autonomous software agents
Daniel Bryant, Surrey Gu · 2007
Autonomous software components, more commonly referred to as software agents, are often cited as a key enabling technology for the next-generation of online services. However, in these services, consisting of large-scale open multi-agent systems, classical forms of reasoning (such as provided by traditional monotonic logics) will be unsuitable. In this work we discuss our vision for a non-monotonic reasoning component for software agents, and present our current work on a prototype light-weight Java-based reasoning engine. The logical machinery underlying our implementation is based on defeasible argumentation, a relatively new paradigm in logical reasoning based upon sound theoretical concepts from the study of argument. The use of argumentation allows an agent to reason about and provide support for opinions, claims, proposals and ultimately decisions and conclusions in a flexible manner. The primary motivations behind this report are to illustrate that a practical agent-based implementation of argumentation is now viable. However, much work remains to be done, particularly in regard to the inherent computational complexity of the process of argumentation. A fundamental message of our work is that we take seriously the need for open source tools and benchmarking applications, which will allow us to gather sound empirical evidence for the applicability of argumentation within this domain.