A New Algebraic Modeling for Distributed Problem-Solving of Multi-Agent Systems(PartI): Social Behavior, Social Situation and Social Dynamics
Shuai Dian · Chinese Journal of Computers · 2002
Basically, there are two categories in distributed artificial intelligence (DAI): distributed problem solving (DPS) and multi agent system (MAS). DSP is concerned with how to increase the whole outcome of the system through cooperations between individual agents regardless of their personal payoffs. Whereas, in MAS each of autonomous rational individuals tries to increase its own personal utility via social interactions, without some global control strategies and overall consistent knowledge, even without some common general goals. These companion papers are devoted to a new algebraic modeling approach to hyper distributed hyper parallel problem solving in MAS, which is entirely different from the DSP and from other methods currently used in MAS. We try to capture the essential characters of social intelligence associated with complicated concurrent social interactions among individual agents in MAS. The algebraic approach universally deals with social interactions by means of social intelligence and social dynamics, and can be effectively used for such problem solving in MAS that might be difficult to solve in the context of social behaviors by using other conventional methods. This is the first of the companion papers, which addresses the typical social behaviours, their essential characters and their formalization. A conceptual architecture of the algebraic modeling approach to MAS problem solving is given. The social situation and social dynamics of algebraic modeling are also discussed. The simulation on distributed self organized multi task allocations and resource assignments shows the advantages of the algebraic approach in the context of social interactions.