Principles and Practice of Multi-Agent Systems

Abstract

This special issue presents four revised and extended papers from PRIMA 2010: the 13th International Conference on Principles and Practice of Multi-Agent Systems, which was held in November 12–15, 2010, in Kolkata, India. Agent computing and technology is an exciting emerging paradigm that is expected to play a key role in many society-changing practices, from disaster response to manufacturing, and from energy management to healthcare. Agent systems are expected to operate in real-world environments, with all the challenges that such environments present. This special issue contains revised and extendedversions of four papers thatwere originally presented at the conference on Principles and Practice of Multi-Agent Systems (PRIMA). PRIMA has been held since 2007, and is successor to a workshop series (also called PRIMA, Pacific Rim International Workshop on Multi-Agents) which has been running since 1998. Each of the four papers in this special issue received high scores from reviewers, and they were given awards at the conference (best paper, two best paper runner-ups, and an IBM Research best paper award for a paper in the area of agents and services). Each of the four papers was invited to be extended, and was then reviewed by the original reviewers, and then checked by the chairs. The four papers cover a wide range of topics, and we hope that they capture the diversity and richness of work in the area, and specifically, of the work that was presented at PRIMA 2010. Effect of DisCSP Variable-Ordering Heuristics in Scale-free Networks, by Tenda Okimoto, Atsushi Iwasaki and Makoto Yokoo. This paper considers the long-standing problem of distributed constraint satisfaction in the case where the network structure of the problem has a particular structure, in this case being scale-free. The authors find that for scale-free networks the choice of variable-ordering heuristic is more significant than in random networks. The authors then develop a heuristic for ordering variables that is especially designed for scale-free networks, and that exhibits improved performance in such networks.