On the Structure of Address in Multi-Party Dialogue: From Discrete Labels to Continuous Levels
Computation and LanguageArtificial IntelligenceHuman-Computer Interaction
Summary
The authors studied how to figure out who a message in a group conversation is meant for. Instead of only picking one person or the whole group, they looked at how messages can be partly meant for multiple people at once. They found that not just who speaks next, but also where people look and how they give small signals (like nods or sounds) relate to who the message is aimed at. Their method, which treats addressing as a gradual thing instead of a simple choice, worked better for predicting these cues. They suggest future research should consider this more flexible way of understanding conversations.
Authors
Taiga Mori, Koji Inoue, Divesh Lala, Tatsuya Kawahara
Abstract
In multi-party dialogues between a dialogue system and multiple users, identifying to whom an utterance is addressed is a key challenge. Prior work has typically treated addressee detection as a multi-class classification task, selecting a single label representing an individual participant or the group. This formulation assumes that address is inherently discrete and has primarily been used for predicting turn-taking. In this paper, we revisit this assumption by analyzing address as a continuous phenomenon. Using a multi-party human dialogue corpus annotated by multiple annotators, we construct both binary address labels derived from majority-vote addressee labels and continuous address levels inferred from annotator judgments using a latent-variable model. We then examine how these representations relate to turn-taking as well as listener behaviors, including gaze and backchannels. Our results show that, in addition to turn-taking, both gaze and backchannels are associated with address. Furthermore, models using continuous address levels achieve better predictive fit than those using discrete labels, suggesting that address may exhibit graded structure. Finally, we discuss the future directions of addressee detection research based on the findings of this study.