Reference-Centric Models for Grounded Collaborative Dialogue

TL;DR

This paper introduces a grounded neural dialogue model that effectively collaborates in a partially observable reference game, significantly improving success rates over previous methods through structured referencing and pragmatic generation.

Contribution

The paper presents a novel reference-centric neural dialogue model that enhances grounding and pragmatic communication in collaborative reference tasks.

Findings

20% improvement in task success in self-play

50% improvement in human evaluation success

Effective grounding of referents in partially observable environments

Abstract

We present a grounded neural dialogue model that successfully collaborates with people in a partially-observable reference game. We focus on a setting where two agents each observe an overlapping part of a world context and need to identify and agree on some object they share. Therefore, the agents should pool their information and communicate pragmatically to solve the task. Our dialogue agent accurately grounds referents from the partner's utterances using a structured reference resolver, conditions on these referents using a recurrent memory, and uses a pragmatic generation procedure to ensure the partner can resolve the references the agent produces. We evaluate on the OneCommon spatial grounding dialogue task (Udagawa and Aizawa 2019), involving a number of dots arranged on a board with continuously varying positions, sizes, and shades. Our agent substantially outperforms the previous state of the art for the task, obtaining a 20% relative improvement in successful task completion in self-play evaluations and a 50% relative improvement in success in human evaluations.