Negotiation is an important skill when interacting actors might have misaligned interests. In order to support automated negotiators research the Automated Negotiation Agent Competition (ANAC) was founded to evaluate automated agents in a bilateral negotiation setting across multiple domains. An analysis of various agents’ strategies from past competitions show that most of them used an explicit opponent modeling component. While it is well known that in repeated interactions, learning the opponent and developing reciprocity become of prominent importance to achieve one’s goal, when the interactions with the same partner are not repeated, focusing on complex opponent modeling might not be the right approach. With that in mind, we explore a domain-based approach in which we form strategies based solely on two domain parameters: the reservation value and the discount factor. Following the presentation of our cognitive model, we present DoNA, a Domain-based Negotiation Agent that exemplifies our approach.
CITATION STYLE
Erez, E. S., & Zuckerman, I. (2016). DoNA—A Domain-based negotiation agent. In Studies in Computational Intelligence (Vol. 638, pp. 261–271). Springer Verlag. https://doi.org/10.1007/978-3-319-30307-9_18
Mendeley helps you to discover research relevant for your work.