Discourse Coherence, Reference Grounding and Goal Oriented Dialogue
This work addresses the problem of improving referential communication in dialogue systems for users, but it is incremental as it builds on existing discourse models and presents initial steps without broad validation.
The paper tackles the problem of mixed-initiative human-computer referential communication by proposing a new approach based on coherence-based discourse models, where utterances attach to an evolving discourse structure and a knowledge graph of speaker commitments interfaces with reasoning and strategy. As a result, they describe a simple dialogue system that accumulates constraints, interprets them with a probabilistic model, and plans clarification using reinforcement learning.
Prior approaches to realizing mixed-initiative human--computer referential communication have adopted information-state or collaborative problem-solving approaches. In this paper, we argue for a new approach, inspired by coherence-based models of discourse such as SDRT \cite{asher-lascarides:2003a}, in which utterances attach to an evolving discourse structure and the associated knowledge graph of speaker commitments serves as an interface to real-world reasoning and conversational strategy. As first steps towards implementing the approach, we describe a simple dialogue system in a referential communication domain that accumulates constraints across discourse, interprets them using a learned probabilistic model, and plans clarification using reinforcement learning.