An infrastructure for rule-driven negotiating software agents

Most strategy-enabled automated negotiation platforms use hard-coded, predefined, and non-adaptive negotiation strategies. This is evidently insufficient in regard to the ambitions of automated negotiations research. In this paper, we introduce INSULA, an infrastructure for representing, managing, and exploring negotiation strategies that is versatile, easy to use, and that encompasses a wide spectrum of negotiation types. We treat negotiation strategies as declarative knowledge, and thus we represent them as if-then rules which soft-ware agents can exploit using an inference engine. The first part of the paper examines negotiation strategies and the possibility of using a rule-based approach to represent them. The second part details the architecture of INSULA, and some of the ongoing experimentations