Abstract
We use reinforcement learning (RL) to learn a multi-issue negotiation dialogue policy. For training and evaluation, we build a hand-crafted agenda-based policy, which serves as the negotiation partner of the RL policy. Both the agendabased and the RL policies are designed to work for a large variety of negotiation settings, and perform well against negotiation partners whose behavior has not been observed before. We evaluate the two models by having them negotiate against each other under various settings. The learned model consistently outperforms the agenda-based model. We also ask human raters to rate negotiation transcripts between the RL policy and the agenda-based policy, regarding the rationality of the two negotiators. The RL policy is perceived as more rational than the agenda-based policy.
Highlights
Negotiation is a process in which two or more parties participate in order to reach a joint decision
We focus on two-party negotiation, and use reinforcement learning (RL) to learn a multi-issue negotiation policy for an agent aimed for negotiating with humans
We have the RL policy interact with the agenda-based simulated user (SU) for 20000 episodes varying the initial settings for both agents in the same fashion as for training
Summary
Negotiation is a process in which two or more parties participate in order to reach a joint decision. Our SU is a hand-crafted negotiation dialogue policy inspired by the agenda paradigm, previously used for dialogue management (Rudnicky and Xu, 1999) and user modeling (Schatzmann and Young, 2009) in information providing tasks Both the agenda-based and the RL policies are designed to work for a variety of goals, preferences, and negotiation moves, even under conditions that are very different from the conditions that the agents have experienced before. Both the agenda-based SU and the RL policy have human-like constraints of imperfect information about each other; they do not know each other’s goals or preferences, number of available arguments, degree of persuadability, or degree of rationality Both agents are required to perform well for a variety of negotiation settings, and against opponents whose negotiation behavior has not been observed before and may vary from one interaction to another or even within the same interaction. To our knowledge this is the first time that RL is used to learn so complex multi-issue negotiation and argumentation policies (how to employ arguments to persuade the other party) designed to work for a large variety of negotiation settings, including settings that did not appear during training
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
