Explanation and Argumentation Capabilities: Towards the Creation of More Persuasive Agents

  • Authors:

  • B. Moulin;H. Irandoust;M. Bé/langer;G. Desbordes

  • Affiliations:

  • Computer Science Department, Laval University, Ste Foy, QC G1K 7P4, Canada&semi/ Research Center on Geomatics, Casault Pavilion, Laval University (E-mail: bernard.moulin@ift.ulaval.cahengameh.irandoust@ift.ulaval.ca< ...;Defence Research Establishment Valcartier, Department of National Defence, Canada (E-mail: micheline.belanger@drev.dnd.ca);Computer Science Department, Laval University, Ste Foy, QC G1K 7P4, Canada

  • Venue:
  • Artificial Intelligence Review
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

During the past two decades many research teamshave worked on the enhancement of theexplanation capabilities of knowledge-basedsystems and decision support systems. Duringthe same period, other researchers have workedon the development of argumentative techniquesfor software systems. We think that it would beinteresting for the researchers belonging tothese different communities to share theirexperiences and to develop systems that takeadvantage of the advances gained in eachdomain.We start by reviewing the evolution ofexplanation systems from the simple reasoningtraces associated with early expert systems torecent research on interactive andcollaborative explanations. We then discuss thecharacteristics of critiquing systems that testthe credibility of the user's solution. Therest of the paper deals with the differentapplication domains that use argumentativetechniques. First, we discuss how argumentativereasoning can be captured by a generalstructure in which a given claim or conclusionis inferred from a set of data and how thisargument structure relates to pragmaticknowledge, explanation production and practicalreasoning. We discuss the role of argument indefeasible reasoning and present some works inthe new field of computer-mediated defeasibleargumentation. We review different applicationdomains such as computer-mediatedcommunication, design rationale, crisismanagement and knowledge management, in whichargumentation support tools are used. Wedescribe models in which arguments areassociated to mental attitudes such as goals,plans and beliefs. We present recent advancesin the application of argumentative techniquesto multi-agent systems. Finally, we proposeresearch perspectives for the integration ofexplanation and argumentation capabilities inknowledge-based systems and make suggestionsfor enhancing the argumentation and persuasioncapabilities of software agents.