DialogXpert: Driving Intelligent and Emotion-Aware Conversations through Online Value-Based Reinforcement Learning with LLM Priors
This addresses the challenge of strategic and emotionally aware dialogue planning for applications like negotiation and tutoring, though it is incremental as it builds on existing LLM and reinforcement learning methods.
The paper tackles the problem of enabling proactive, goal-driven conversations in LLM agents by introducing DialogXpert, which uses a frozen LLM for candidate actions and a Q-network for selection, achieving success rates over 94% in under 3 turns across benchmarks.
Large-language-model (LLM) agents excel at reactive dialogue but struggle with proactive, goal-driven interactions due to myopic decoding and costly planning. We introduce DialogXpert, which leverages a frozen LLM to propose a small, high-quality set of candidate actions per turn and employs a compact Q-network over fixed BERT embeddings trained via temporal-difference learning to select optimal moves within this reduced space. By tracking the user's emotions, DialogXpert tailors each decision to advance the task while nurturing a genuine, empathetic connection. Across negotiation, emotional support, and tutoring benchmarks, DialogXpert drives conversations to under $3$ turns with success rates exceeding 94\% and, with a larger LLM prior, pushes success above 97\% while markedly improving negotiation outcomes. This framework delivers real-time, strategic, and emotionally intelligent dialogue planning at scale. Code available at https://github.com/declare-lab/dialogxpert/