Identifying Culture and Leveraging Cultural Differences for Negotiation Agents

Enabling computational agents to efficiently aid or automate negotiations with humans requires a recognition and understanding of differences in negotiation behavior. One important dimension of these differences is based on the cultures of the negotiators. In this work, we investigate two tasks needed to enable culturally-sensitive computational negotiation agents: (1) the identification of negotiators' cultures from observed negotiation transcripts, and (2) the prediction of culturally-based differences in negotiation outcomes. We employ negotiation data collected from negotiators of varying cultures in a multi-issue negotiation scenario to investigate these tasks.