LEEETs-Dial: Linguistic Entrainment in End-to-End Task-oriented Dialogue systems
This addresses the problem of unnatural dialogue interactions for users in task-oriented systems, though it is incremental as it builds on existing GPT-2 models.
The paper tackled the problem of linguistic entrainment (alignment) in task-oriented dialogue systems, which improves user experience but is often lacking. The result showed that three GPT-2-based methods—training instance weighting, entrainment-specific loss, and additional conditioning—significantly outperformed a base model in entrainment, as confirmed by automated and manual metrics.
Linguistic entrainment, or alignment, represents a phenomenon where linguistic patterns employed by conversational participants converge to one another. While entrainment has been shown to produce a more natural user experience, most dialogue systems do not have any provisions for it. In this work, we introduce methods for achieving dialogue entrainment in a GPT-2-based end-to-end task-oriented dialogue system through the utilization of shared vocabulary. We experiment with training instance weighting, entrainment-specific loss, and additional conditioning to generate responses that align with the user. We demonstrate that all three approaches produce significantly better entrainment than the base, non-entrainment-optimized model, as confirmed by both automated and manual evaluation metrics.