ProDial – an annotated proactive dialogue act corpus for conversational assistants using crowdsourcing

  • Proactive behaviour is an integral interaction concept of both human-human as well as human-computer cooperation. However, modelling proactive systems and appropriate interaction strategies are still an open quest. In this work, a parameterised and annotated dialogue corpus has been created. The corpus is based on human interactions with an autonomous agent embedded in a serious game setting. For modelling proactive dialogue behaviour, the agent was capable of selecting from four different proactive actions (None, Notification, Suggestion, Intervention) in order to serve as the user’s personal advisor in a sequential planning task. Data was collected online using crowdsourcing (308 participants) resulting in a total of 3696 system-user exchanges. Data was annotated with objective features as well as subjectively self-reported features for capturing the interplay between proactive behaviour and situational as well as user-dependent characteristics. The corpus is intended for building aProactive behaviour is an integral interaction concept of both human-human as well as human-computer cooperation. However, modelling proactive systems and appropriate interaction strategies are still an open quest. In this work, a parameterised and annotated dialogue corpus has been created. The corpus is based on human interactions with an autonomous agent embedded in a serious game setting. For modelling proactive dialogue behaviour, the agent was capable of selecting from four different proactive actions (None, Notification, Suggestion, Intervention) in order to serve as the user’s personal advisor in a sequential planning task. Data was collected online using crowdsourcing (308 participants) resulting in a total of 3696 system-user exchanges. Data was annotated with objective features as well as subjectively self-reported features for capturing the interplay between proactive behaviour and situational as well as user-dependent characteristics. The corpus is intended for building a user model for developing trustworthy proactive interaction strategies.show moreshow less

Download full text files

Export metadata

Statistics

Number of document requests

Additional Services

Share in Twitter Search Google Scholar
Metadaten
Author:Matthias KrausORCiDGND, Nicolas Wagner, Wolfgang Minker
URN:urn:nbn:de:bvb:384-opus4-1013696
Frontdoor URLhttps://opus.bibliothek.uni-augsburg.de/opus4/101369
URL:https://aclanthology.org/2022.lrec-1.339
ISBN:979-10-95546-72-6OPAC
Parent Title (English):Proceedings of the Thirteenth Language Resources and Evaluation Conference (LREC), 20-25 June 2022, Marseille, France
Publisher:European Language Resources Association (ELRA)
Place of publication:Paris
Editor:Nicoletta Calzolari, Frédéric Béchet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hitoshi Isahara, Bente Maegaard, Joseph Mariani, Hélène Mazo, Jan Odijk, Stelios Peperidis
Type:Conference Proceeding
Language:English
Year of first Publication:2022
Publishing Institution:Universität Augsburg
Release Date:2023/01/30
First Page:3164
Last Page:3173
Institutes:Fakultät für Angewandte Informatik
Fakultät für Angewandte Informatik / Institut für Informatik
Fakultät für Angewandte Informatik / Institut für Informatik / Lehrstuhl für Menschzentrierte Künstliche Intelligenz
Dewey Decimal Classification:0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 004 Datenverarbeitung; Informatik
Licence (German):CC-BY-NC 4.0: Creative Commons: Namensnennung - Nicht kommerziell (mit Print on Demand)