BABE Ş - BOLYAI UNIVERSITY FACULTY OF ECONOMICS BUSINESS INFORMATION SYSTEMS DEPARTMENT CONTRIBUTIONS TO CONCEPTION , DESIGN AND DEVELOPMENT OF COLLABORATIVE MULTI - AGENT SYSTEMS PhD candidate : Gheorghe Cosmin Silaghi

Programming has progressed through machine code, assembly language, machine-independent programming languages, procedures and functions, abstract data types and objects to a new paradigm: agents. A new field emerged in Artificial Intelligence: multi-agent systems. Intelligent agents became the most vibrant and fastest growing research area in both Artificial Intelligence and Computer Science. Multi-agent systems represent a modern approach for modeling so-called open systems. They represent a new modeling technique and a simulation paradigm for both social and economic organizations and for software conception and design. Therefore, the study of how we can build multi-agent systems becomes of great interest. This Ph.D. thesis focuses on multi-agent systems from the point of view of enabling technologies: concepts, methods and tools. We investigate, as a first study objective, collaboration fundamentals, as being the foundation on which one should design new agent societies of every kind, human societies or artificial ones. We approach the theoretical tools for formalizing interaction, in general and soft-achieved collaboration, in particular and we exemplify these tools on some remarkable theoretical problems. Our intention is to evaluate these tools and to extract some remarks about when and how they are worth to be used. Settings like voting, auctions, bargaining, market mechanisms and coalition formation are investigated. As the thesis is at the border between computer science and economics, we emphasize on some computational issues an economist usually does not take into consideration. The second study objective of the thesis is to find out if agent technology is mature enough and if it can be adopted by industry. Therefore, we chose to investigate some design and development techniques for building agent systems. Beside a theoretical approach, we present the way a multi-agent based system can solve a complex problem: web searching. Web information retrieval domain is complex enough to pose multiple design choices and to force us to organize our software project based on productivity principles. A collaborative solution is considered for the web search problem, in the idea that agent interaction can give out a better personalized retrieval than pure machine-based approach. This topic is valuable to test both the design and development of agent systems and collaborative interactions of agents, in an environment where intelligent and hard-to-run algorithms need to be employed.