Fuzzy referral based cooperation in social networks of agents

Since intelligent agents aim to represent human reasoning and behaviour, multiagent systems are supposed to reflect the collective behaviour of human societies. There, each individual determine who collaborate with, forming a social network to improve the quality of the decisions to be made. Several models of computation have been proposed in literature. Among them, the approach of the authors relies on the ability of fuzzy logic to operate with vague and uncertain terms. This paper studies how much effective is the cooperation between agents modelled by these proposals in different scenarios. The scenarios test the evolution of the deception caused by overvalued/undervalued predictions, with cooperation of low/high reliability, and with three different levels of variability in the behaviour of the agents evaluated.

[1]  José M. Molina López,et al.  A BDI agent architecture for reasoning about reputation , 2001, 2001 IEEE International Conference on Systems, Man and Cybernetics. e-Systems and e-Man for Cybernetics in Cyberspace (Cat.No.01CH37236).

[2]  David W. McDonald,et al.  Recommending collaboration with social networks , 2003, CHI 2003.

[3]  Munindar P. Singh,et al.  A Social Mechanism of Reputation Management in Electronic Communities , 2000, CIA.

[4]  Bart Selman,et al.  Referral Web: combining social networks and collaborative filtering , 1997, CACM.

[5]  José M. Molina López,et al.  Trust Management Through Fuzzy Reputation , 2003, International Journal of Cooperative Information Systems.

[6]  José M. Molina López,et al.  A fuzzy model of reputation in multi-agent systems , 2001, AGENTS '01.

[7]  Stephen Hailes,et al.  Supporting trust in virtual communities , 2000, Proceedings of the 33rd Annual Hawaii International Conference on System Sciences.

[8]  Jordi Sabater-Mir,et al.  Reputation and social network analysis in multi-agent systems , 2002, AAMAS '02.