Offline Norm Evolution
2015 (English)In: Agents and Artificial Intelligence: 7th International Conference, ICAART 2015 Lisbon, Portugal, January 10–12, 2015 Revised Selected Papers / [ed] Béatrice Duval, Jaap van den Herik, Stephane Loiseau, Joaquim Filipe, Cham: Springer, 2015, 316-333 p.Conference paper (Refereed)
An approach to the pre-runtime design of normative systems for a class of problem-solving norm-regulated multi-agent systems is suggested. The basic idea is to employ evolutionary mechanisms to evolve efficient normative systems for so-called norm-regulated Dalmases, as part of the design process. The Dalmas architecture uses an algebraic approach to normative systems, in which normative consequences are based on an extended set of one-agent types of normative positions, which is given a semantics in terms of prohibition of certain types of state transitions. To illustrate the approach, a genetic algorithm is used to evolve norms for an example system. Furthermore, some approaches to reducing the algorithm's search space, including to employ a notion of operational equivalence of norms, are discussed. It is demonstrated that an evolutionary algorithm may be a useful tool when designing norms for problem-solving multi-agent systems.
Place, publisher, year, edition, pages
Cham: Springer, 2015. 316-333 p.
, Lecture notes in artificial intelligence, ISSN 0302-9743 ; 9494
Norm-regulated Multi-Agent System, Normative MAS, DALMAS, Norm Evolution, Evolutionary Algorithm
Research subject Computer and Systems Sciences
IdentifiersURN: urn:nbn:se:su:diva-120329DOI: 10.1007/978-3-319-27947-3_17ScopusID: 2-s2.0-84952784465ISBN: 978-331927946-6ISBN: 978-3-319-27947-3OAI: oai:DiVA.org:su-120329DiVA: diva2:851675
The 7th International Conference on Agents and Artificial Intelligence, ICAART 2015, Lisbon, Portugal, 10-12 January, 2015