Opponent Modeling in Negotiation Dialogues by Related Data Adaptation

Anonymous

Abstract: Opponent modeling refers to the task of inferring another party's mental state within the context of non-collaborative social tasks. In a negotiation, it involves identifying the opponent’s priorities, which is crucial for finding high-value deals. Discovering these priorities is helpful for automated negotiation systems deployed in pedagogy and conversational AI. In this work, we propose a transformer-based ranker for identifying these priorities from negotiation dialogues. The model takes in a partial dialogue as input and predicts the priority order of the opponent. We further devise ways to adapt related data sources for this task to provide more explicit supervision for incorporating the opponent preferences and offers, as a proxy to relying on granular utterance-level annotations. We show the utility of our proposed approach through extensive experiments based on two dialogue datasets. We particularly find that the proposed data adaptations lead to strong performance in 0-shot and few-shot scenarios. Moreover, they allow the model to perform better with access to fewer utterances from the opponent.

Paper Type: long