The Interplay of Task Success and Dialogue Quality: An in-depth Evaluation in Task-Oriented Visual Dialogues

Alberto Testoni, Raffaella Bernardi


Abstract
When training a model on referential dialogue guessing games, the best model is usually chosen based on its task success. We show that in the popular end-to-end approach, this choice prevents the model from learning to generate linguistically richer dialogues, since the acquisition of language proficiency takes longer than learning the guessing task. By comparing models playing different games (GuessWhat, GuessWhich, and Mutual Friends), we show that this discrepancy is model- and task-agnostic. We investigate whether and when better language quality could lead to higher task success. We show that in GuessWhat, models could increase their accuracy if they learn to ground, encode, and decode also words that do not occur frequently in the training set.
Anthology ID:
2021.eacl-main.178
Volume:
Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume
Month:
April
Year:
2021
Address:
Online
Editors:
Paola Merlo, Jorg Tiedemann, Reut Tsarfaty
Venue:
EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2071–2082
Language:
URL:
https://aclanthology.org/2021.eacl-main.178
DOI:
10.18653/v1/2021.eacl-main.178
Bibkey:
Cite (ACL):
Alberto Testoni and Raffaella Bernardi. 2021. The Interplay of Task Success and Dialogue Quality: An in-depth Evaluation in Task-Oriented Visual Dialogues. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 2071–2082, Online. Association for Computational Linguistics.
Cite (Informal):
The Interplay of Task Success and Dialogue Quality: An in-depth Evaluation in Task-Oriented Visual Dialogues (Testoni & Bernardi, EACL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.eacl-main.178.pdf
Code
 stanfordnlp/cocoa
Data
GuessWhat?!MutualFriendsVisDial