Semantic-Aware Action Space Compression via LLM-DRL Synergy for Efficient Task-oriented Dialogue Policy Exploration

ACL ARR 2025 May Submission3639 Authors

Abstract: The flexibility of natural language significantly expands the action space in task-oriented dialogue systems, causing inefficient exploration and slow convergence in deep reinforcement learning (DRL)-based policy optimization. Pre-trained large language models (LLMs), with world knowledge and semantic understanding, offer promising solutions. To this end, we propose LLM-Guided DRL via Semantic-Aware Action Pruning (LLMSAP), a novel framework that synergizes pretrained LLMs with DRL. LLMSAP leverages the world knowledge and contextual understanding of LLMs to guide decision-making via an action feasibility assessment. Instead of requiring LLMs to directly generate optimal actions due to their limited precision in sequential decision tasks, LLMSAP employs a lightweight action pruning mechanism. Specifically, LLMs act as action filters, rapidly eliminating semantically implausible or low-potential actions from multi-turn dialogue context, allowing the DRL agent to focus exploration on a refined candidate subset. This two-stage framework ("prune-then-optimize") avoids extensive LLM fine-tuning while preserving the decision-making precision of DRL. Experiments on multiple benchmarks verify the effectiveness of LLMSAP.

Paper Type: Short

Research Area: Dialogue and Interactive Systems

Research Area Keywords: Task-Oriented Dialogue System, Deep Reinforcement Learning, Large Language Model, Dialogue Policy

Contribution Types: NLP engineering experiment

Languages Studied: English

Keywords: Task-Oriented Dialogue System, Deep Reinforcement Learning, Large Language Model, Dialogue Policy

Submission Number: 3639