| Abstract |
Large ontologies and semantic networks represent complex multilevel structures, which are incredibly resistant to standard proof checking procedures. Automatic consistency checks can discover system errors such as missing intralingual links, but to find a missing word sense is a difficult task. Standard solutions rely on successive consultations of multiple information sources in a multi-level review process. In this paper, we present a new approach of supplementing such multi-level reviews with engaging the dictionary users in WordNet error corrections and enhancement proposals via systematic crowdsourcing. This approach defines an early release phase with the full dataset published to the target audience followed by a continuous workflow consisting of structured adjustment suggestions obtained from the public users and of the complete editing process by expert reviewers. The review team members are handling prestructured review tasks organized in aggregated forms with correction proposals, the revision management and the appropriate editing of proposed changes. Both the users and reviewers have access to the complete revision history, which allows them to handle repeated proposals responsibly. |
| BibTex |
@InProceedings{ELX2018-091,
author={Aleš Horák, Adam Rambousek},
title={Wordnet Consistency Checking via Crowdsourcing},
pages={1023-1029},
booktitle={Proceedings of the XVIII EURALEX International Congress: Lexicography in Global Contexts},
year={2018},
month={jul},
date={17-21},
address={Ljubljana, Slovenia},
editor={Jaka Čibej, Vojko Gorjanc, Iztok Kosem, Simon Krek},
publisher={Ljubljana University Press, Faculty of Arts},
isbn={978-961-06-0097-8}, } |
