CoDial: Interpretable Task-Oriented Dialogue Systems Through Dialogue Flow Alignment
🤖 AI Summary
Building task-oriented dialogue systems for high-stakes domains (e.g., law, healthcare) remains challenging for non-technical domain experts due to the difficulty of low-threshold definition, debugging, and optimization of dialogue behaviors.
Method: We propose a novel framework featuring (1) a structured heterogeneous graph that explicitly encodes domain expert knowledge, enabling zero-shot, interpretable, and editable dialogue policy generation; and (2) a human-in-the-loop, LLM-assisted iterative alignment mechanism that supports efficient, annotation-free, and retraining-free optimization.
Contribution/Results: Evaluated on STAR and MultiWOZ, our framework achieves state-of-the-art performance among inference-only models on STAR and matches mainstream baselines on MultiWOZ. It significantly enhances expert-driven iterative development efficiency and controllability—enabling rapid, transparent, and precise customization of dialogue logic without ML expertise.
Application Category
📝 Abstract
It is often challenging to teach specialized, unseen tasks to dialogue systems due to the high cost of expert knowledge, training data, and high technical difficulty. To support domain-specific applications - such as law, medicine, or finance - it is essential to build frameworks that enable non-technical experts to define, test, and refine system behaviour with minimal effort. Achieving this requires cross-disciplinary collaboration between developers and domain specialists. In this work, we introduce a novel framework, CoDial (Code for Dialogue), that converts expert knowledge, represented as a novel structured heterogeneous graph, into executable conversation logic. CoDial can be easily implemented in existing guardrailing languages, such as Colang, to enable interpretable, modifiable, and true zero-shot specification of task-oriented dialogue systems. Empirically, CoDial achieves state-of-the-art performance on the STAR dataset for inference-based models and is competitive with similar baselines on the well-known MultiWOZ dataset. We also demonstrate CoDial's iterative improvement via manual and LLM-aided feedback, making it a practical tool for expert-guided alignment of LLMs in high-stakes domains.
Problem
Research questions and friction points this paper is trying to address.
Teaching specialized tasks to dialogue systems is costly and technically difficult
Non-technical experts need frameworks to define and refine system behavior easily
Cross-disciplinary collaboration is required for domain-specific dialogue system development
Innovation
Methods, ideas, or system contributions that make the work stand out.
Converts expert knowledge into executable logic
Uses structured heterogeneous graph representation
Integrates with guardrailing languages like Colang
🔎 Similar Papers
No similar papers found.
