Imperial College London
Universita' di Bologna
The main purpose of the CLIMA workshop series is the promotion of research at the intersection of Computational Logic and Multi-Agent Systems.
The series initiated in affiliation to major Computational Logic and AI events (past editions in Portugal, USA, Cyprus, Denmark, and the United Kingdom), and this year has been run for the first time as a stand-alone event.
The "Sixth International Workshop on Computational Logic in Multi-Agent Systems" (CLIMA VI) was held at City University London, UK, from 27th to 29th of June 2005. The event was characterised by a strong emphasis on dissemination: research done in recent years in this area, and the success of last year's edition, encouraged us to consider CLIMA as a golden opportunity for broadening this research community, and for strengthening the role of CLIMA itself as one of its reference events.
For this reason, we decided to give the workshop a novel shape. In addition to the usual technical programme (16 high quality papers selected from 30 submissions) and the invited talk by Bob Kowalski, we proposed a tutorial programme, a dissemination event for SOCS, a EU-funded project on agents and computational logic, and a contest. Our endeavour was rewarded by an impressive response, with about 60 delegates coming from Canada, Cyprus, France, Germany, Ireland, Italy, Japan, Lithuania, the Netherlands, Norway, Poland, Spain,
Sweden, Switzerland, and the United Kingdom. Over a third of the audience was composed by research students, who found in CLIMA an occasion to get to know one another, establish contacts and exchange views on their research interests. This record attendance would not have been possible without sponsorship: we especially thank the support provided by the SOCS project, as well as that coming from AgentLink and from the Association for Logic Programming, which was entirely devoted to financing the participation of nearly 20 delegates worldwide.
The tutorial programme covered a broad range of topics: from rational agent programming to normative and social reasoning and verification, and a number of formalisms, from temporal logics to logic programming to deontic logic and beyond.
The aims of this programme were to introduce Computational Logic-based agent programming environments, to introduce novices to state-of-the-art Computational Logic-based research in Multi-Agent Systems, and to provide guidelines to researchers and practitioners interested in logic-based agent technologies. Invited lectures have been given by Rafael Bordini on BDI agent programming in AgentSpeak, through an overview of the features available with the agent programming language Jason; by Michael Fisher, on rational agent groups programming using a combination of temporal logic and logics concerning belief and ability; by Keith Clark and Silvana Zappacosta on multi-agent systems programming based on the multi-threaded and distributed logic programming framework Qu-Prolog; by Marek Sergot on the representation of norms of behaviour and institutional aspects of (human or computer) societies, using the language (C+)++; by Fariba Sadri and Kostas Stathis on the KGP model of agency and how to use abductive logic programming and logic programming with priorities for agent specification and programming; and by Federico Chesani and Marco Gavanelli on designing, specifying, verifying and testing interaction protocols for open agent societies, using SOCS-SI.
The EU-funded SOCS project dissemination event consisted of two tutorials (on the KGP model and on the SOCS-SI framework, both outcomes of SOCS), and of a number of talks, aimed at presenting and discussing the main research challenges and achievements of the project. SOCS (IST-2001-32530, http://lia.deis.unibo.it/research/socs/) is a 42 months EU-funded research project, ended June 2005, which involved 6European academic partners and whose purpose was to produce a Computational Logic model for the description, analysis and verification of global and open societies of heterogeneous agents, called computees (standing for computational logic-based agents). The talks have been given by Francesca Toni (project co-ordinator) and by Marco Alberti, Andrea Bracciali, Tony Kakas, and Paolo Torroni. Among the topics covered by such an event: the motivations to adopt formal approaches to programming Multi-Agent Systems, the kinds of results to be expected from such research, the problems of combining different Computational Logic frameworks into a single agent framework, and finally the issues of evaluating and testing rational agent systems.
The purpose of the First CLIMA Contest was to establish evaluation criteria that can serve as milestones for testing new approaches and techniques. The contest as an instrument to evaluate and foster research has been successfully used in various other parts of artificial intelligence (theorem proving, planning, operations research, constraint programming, robo-cup etc.) and, lately, also in specialised areas in agent systems (trading agents). We decided to promote a CLIMA contest, where original, innovative, and effective application of Computational Logic-based techniques can be
confronted in solving specific multi-agent issues. The contest was organised by Mehdi Dastani and Juergen Dix. In order to render the contest concrete, the organisers opted for a particular scenario that served as a basis for such a contest. The formulation of such a scenario has turned out to be a very hard task. The contest has been one of the most entertaining moments of CLIMA VI. The winners of this first edition have been two teams, which both ranked first: Simon Coffey and Dorian Gaertner, from Imperial College London, UK, and Carlos Cares, Xavier Franch and Enric Mayol, from Universitat Politecnica de Catalunya, Barcelona, Spain,and Universidad de la Frontera, Temuco, Chile. We hope that this initiative will be a first step towards the definition of paradigms and measuring standards that can further logic-based intelligent agent research and its application.
The CLIMA VI programme was completed by the invited talk by the CLIMA VI keynote speaker, Bob Kowalski, who advocated the need of decision theoretic features in logic-based agents, and motivated the need for a pre-activity feature in agent systems, in addition to the conventional reactivity and pro-activity features.
Overall, we were extremely happy with the success of the event, and our impression has been backed up by a very positive feedback given by several delegates. The sessions have been in fact very crowded and lively throughout the whole duration of the workshop - in spite of a very demanding programme, which included a London quest, in which a crowd of heroic delegates managed to find the venue of the social dinner at the OXO tower and a WTF funky concert after the dinner.
While the post-proceedings of the past CLIMA V event are now being published by Springer as LNAI 3487, we are now organizing the edition of a CLIMA VI book, to be published in the same series by Springer, with a selection of the articles presented at the workshop, and a number of invited contribution.
CLIMA VII will be held in Japan. It will be organized by Katsumi Inoue, Ken Satoh and Francesca Toni in coordination with the Fifth AAMAS conference, and it will host the second CLIMA contest.