vix.ing · top · new · best · stats · spec

The PhotoBook Dataset: Building Common Ground through Visually-Grounded\n Dialogue

2019/06/04 by Janosch Haber, Haber, Janosch, Tim Baumgärtner +9 · 3 citations
Computer Science · Psychology · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Language, Metaphor, and Cognition #Multimodal Machine Learning Applications #Speech and dialogue systems

paper · pdf · doi:10.48550/arxiv.1906.01530

openalex publication_date 2019/06/04 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper introduces the PhotoBook dataset, a large-scale collection of\nvisually-grounded, task-oriented dialogues in English designed to investigate\nshared dialogue history accumulating during conversation. Taking inspiration\nfrom seminal work on dialogue analysis, we propose a data-collection task\nformulated as a collaborative game prompting two online participants to refer\nto images utilising both their visual context as well as previously established\nreferring expressions. We provide a detailed description of the task setup and\na thorough analysis of the 2,500 dialogues collected. To further illustrate the\nnovel features of the dataset, we propose a baseline model for reference\nresolution which uses a simple method to take into account shared information\naccumulated in a reference chain. Our results show that this information is\nparticularly important to resolve later descriptions and underline the need to\ndevelop more sophisticated models of common ground in dialogue interaction.\n

Cited by

Related