Crowd-sourced collection of task-oriented human-human dialogues in a multi-domain scenario
File(s)
Accepted version
Author(s)
Braunschweiler, Norbert
Papadakos, Panagiotis
Kotti, Margarita
Marketakis, Yannis
Tzitzikas, Yannis
Type
Conference Paper
Abstract
There is a lack of high-quality corpora for the purposes of trainingtask-oriented, end-to-end dialogue systems. This paper describes a dialogue col-lection process which used crowd-sourcing and a Wizard-of-Oz set-up to collectwritten human-human dialogues for a task-oriented, multi-domain scenario. Thecontext is a tourism agency, where users try to select the more desired hotel,restaurant, museum or shop. To respond to users, wizards were assisted by an ex-ploratory system supporting Preference-enriched Faceted Search. An importantaspect was the translation of user intent to a number of actions (hard or soft-constraints) by wizards. The main goal was to collect dialogues as realistic aspossible between a user and an operator, suitable for training end-to-end dialoguesystems. This work describes the experiences made, the options and the deci-sions taken to minimize the human effort and budget, along with the tools usedand developed, and describes in detail the resulting dialogue collection.
Date Issued
2019-08-06
Date Acceptance
2019-05-03
Citation
Lecture Notes in Computer Science, 2019, 11697, pp.398-411
ISSN
0302-9743
Publisher
Springer Verlag
Start Page
398
End Page
411
Journal / Book Title
Lecture Notes in Computer Science
Volume
11697
Copyright Statement
© Springer Nature Switzerland AG 2019. The final publication is available at Springer via https://link.springer.com/chapter/10.1007%2F978-3-030-27947-9_34
Source
22th International Conference of Text, Speech and Dialogue (TSD 2019)
Subjects
Artificial Intelligence & Image Processing
Publication Status
Published
Start Date
2019-09-10
Finish Date
2019-09-13
Coverage Spatial
Ljubljana, Slovenia
Date Publish Online
2019-08-06