Toward BDI Sapient Agents: Learning Intentionally

Sapient agents have been characterized as systems that learn their cognitive state and capabilities through experience, considering social environments and interactions with other agents or humans. The BDI (belief, desire, intention) model of cognitive agency offers philosophical grounds on intentionality and practical reasoning, as well as an elegant abstract logical semantics. However, the lack of learning and social competences of this model constitutes a serious limitation when sapient agents are the issue. This chapter discusses some ideas on intentional and social learning, that support the revision of practical reasons by the BDI agents. The resulting agents can learn, and then update, their plans’ contexts to avoid forming intentions that eventually fail. Individual and social learning have been successfully attained in our own BDI interpreter based on dMARS, and ported to the Jason interpreter based on AgentSpeak(L). Multiagent consistency is guaranteed through a protocol based on cooperative goal adoption. These BDI learning agents seems closer to the intended characterization of sapient agents.

[1]  Mehdi Dastani,et al.  A characterization of sapient agents , 2003, IEMC '03 Proceedings. Managing Technologically Driven Organizations: The Human Side of Innovation and Change (IEEE Cat. No.03CH37502).

[2]  Cristiano Castelfranchi,et al.  Artificial Intelligence Modelling social action for AI agents , 2003 .

[3]  Amal El Fallah Seghrouchni,et al.  On learning intentionally , 2005, Inteligencia Artif..

[4]  Michael E. Bratman,et al.  Intention, Plans, and Practical Reason , 1991 .

[5]  Rafael H. Bordini,et al.  BDI agent programming in AgentSpeak using Jason , 2006 .

[6]  D. Dennett.  The Intentional Stance. , 1987 .

[7]  Michael Wooldridge,et al.  Reasoning about rational agents , 2000, Intelligent robots and autonomous agents.

[8]  Michael Wooldridge,et al.  The Belief-Desire-Intention Model of Agency , 1998, ATAL.

[9]  Amal El Fallah Seghrouchni,et al.  Learning in BDI Multi-agent Systems , 2004, CLIMA.

[10]  A. S. Roa,et al.  AgentSpeak(L): BDI agents speak out in a logical computable language , 1996 .

[11]  Gerhard Weiss,et al.  Multiagent systems: a modern approach to distributed artificial intelligence , 1999 .

[12]  Amal El Fallah Seghrouchni,et al.  SMILE: Sound Multi-agent Incremental LEarning , 2007, AAMAS '07.

[13]  Michael Wooldridge,et al.  A Formal Specification of dMARS , 1997, ATAL.

[14]  Peter Norvig,et al.  Artificial Intelligence: A Modern Approach , 1995 .

[15]  Anand S. Rao,et al.  Decision Procedures for BDI Logics , 1998, J. Log. Comput..

[16]  Amal El Fallah Seghrouchni,et al.  Distributed learning in intentional BDI multi-agent systems , 2004, Proceedings of the Fifth Mexican International Conference in Computer Science, 2004. ENC 2004..

[17]  David Lightfoot.  Formal Specification Using Z , 1991 .

[18]  Luc De Raedt,et al.  Inductive Logic Programming: Theory and Methods , 1994, J. Log. Program..

[19]  Hendrik Blockeel,et al.  Top-Down Induction of First Order Logical Decision Trees , 1998, AI Commun..

[20]  Munindar P. Singh,et al.  Formal methods in DAI: logic-based representation and reasoning , 1999 .

[21]  Bart Demoen,et al.  Executing Query Packs in ILP , 2000, ILP.

[22]  Henry Soldano,et al.  BDI MULTIAGENT LEARNING BASED ON FIRST-ORDER INDUCTION OF LOGICAL DECISION TREES , 2001 .

[23]  J. Ross Quinlan,et al.  Induction of Decision Trees , 1986, Machine Learning.