Supervised Disambiguation of German Verbal Idioms with a BiLSTM Architecture

Rafael Ehren, Timm Lichte, Laura Kallmeyer, Jakub Waszczuk


Abstract
Supervised disambiguation of verbal idioms (VID) poses special demands on the quality and quantity of the annotated data used for learning and evaluation. In this paper, we present a new VID corpus for German and perform a series of VID disambiguation experiments on it. Our best classifier, based on a neural architecture, yields an error reduction across VIDs of 57% in terms of accuracy compared to a simple majority baseline.
Anthology ID:
2020.figlang-1.29
Volume:
Proceedings of the Second Workshop on Figurative Language Processing
Month:
July
Year:
2020
Address:
Online
Venue:
Fig-Lang
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
211–220
Language:
URL:
https://aclanthology.org/2020.figlang-1.29
DOI:
10.18653/v1/2020.figlang-1.29
Bibkey:
Cite (ACL):
Rafael Ehren, Timm Lichte, Laura Kallmeyer, and Jakub Waszczuk. 2020. Supervised Disambiguation of German Verbal Idioms with a BiLSTM Architecture. In Proceedings of the Second Workshop on Figurative Language Processing, pages 211–220, Online. Association for Computational Linguistics.
Cite (Informal):
Supervised Disambiguation of German Verbal Idioms with a BiLSTM Architecture (Ehren et al., Fig-Lang 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.figlang-1.29.pdf
Video:
 http://slideslive.com/38929715