Schema-Guided Response Generation using Multi-Frame Dialogue State for Motivational Interviewing Systems

2025-08-28

AI Summary

This study addresses the challenge of ensuring adherence to Motivational Interviewing (MI) principles in large language model (LLM)-driven counseling dialogue systems. Methodologically, we propose a framework integrating multi-turn dialogue state modeling with dynamic response focus control. It features a fine-grained, MI-principle-guided dialogue state update mechanism, pattern-informed dialogue management, strategy-controllable LLM response generation, and principle-constrained dynamic focus modulation. Our key contribution is the first explicit computational encoding of MI’s clinical logic—formalizing state transitions and strategy selection rules to precisely guide autonomy support, empathic responding, and evocative questioning. A user study demonstrates that our system significantly improves MI fidelity (+32.7%) and enhances users’ depth of self-reflection and readiness for behavior change (p < 0.01).

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The primary goal of Motivational Interviewing (MI) is to help clients build their own motivation for behavioral change. To support this in dialogue systems, it is essential to guide large language models (LLMs) to generate counselor responses aligned with MI principles. By employing a schema-guided approach, this study proposes a method for updating multi-frame dialogue states and a strategy decision mechanism that dynamically determines the response focus in a manner grounded in MI principles. The proposed method was implemented in a dialogue system and evaluated through a user study. Results showed that the proposed system successfully generated MI-favorable responses and effectively encouraged the user's (client's) deliberation by asking eliciting questions.

Problem

• Guide LLMs to generate MI-aligned counselor responses
• Update multi-frame dialogue states using schema-guided approach
• Dynamically determine response focus with MI principles

Innovation

• Schema-guided response generation with multi-frame dialogue states
• Dynamic strategy decision mechanism for MI principles
• Generating MI-favorable responses through eliciting questions