@inproceedings { , title = {The Task2Dial Dataset: A Novel Dataset for Commonsense-enhanced Task-based Dialogue Grounded in Documents}, abstract = {This paper describes the Task2Dial dataset, a novel dataset of document-grounded task-based dialogues in the food preparation domain , where an Information Giver (IG) provides instructions to an Information Follower (IF) so that the latter can successfully complete the task. In this novel setting, the IF can ask clarification questions which might not be able to be grounded in the underlying document and might require commonsense knowledge to be answered. The Task2Dial dataset poses new challenges: (1) its human reference texts show more lexical richness and variation than other document-grounded dialogue datasets; (2) generating from this set requires paraphrasing as instructional responses have been modified from the underlying recipe; (3) and commonsense knowledge, since questions might not necessarily be grounded in the document ; (4) generating requires planning based on context, as recipe steps need to be provided in order. As such, learning from this dataset promises more natural, varied and less template-like system utterances. The dataset contains dialogues with an average 18.15 number of turns and 19.79 tokens per turn, as compared to 12.94 and 12 respectively in existing datasets. Finally, we also provide a data statement , and we discuss the challenges associated with this novel task/dataset.}, conference = {4th International Conference on Natural Language and Speech Processing (ICNLSP 2021)}, pages = {242-251}, publicationstatus = {Published}, publisher = {Association for Computational Linguistics (ACL)}, url = {http://researchrepository.napier.ac.uk/Output/2807732}, year = {2024}, author = {Strathearn, Carl and Gkatzia, Dimitra} }