Code and data for the COLING 2018 paper "Killing Four Birds with Two Stones: Multi-Task Learning for Non-Literal Language Detection" by Erik-Lân Do Dinh, Steffen Eger, and Iryna Gurevych. Paper PDF available at ACL Anthology: https://aclweb.org/anthology/C18-1132
Please use the following citation:
@InProceedings{dodinh:2018:coling,
author = {Do Dinh, Erik-Lân and Eger, Steffen and Gurevych, Iryna},
title = {Killing Four Birds with Two Stones: Multi-Task Learning for Non-Literal Language Detection},
booktitle = {Proceedings of the 27th International Conference on Computational Linguistics},
month = aug,
year = {2018},
address = {Santa Fe, NM, USA},
publisher = {The COLING 2018 Organizing Committee},
pages = {1558--1569},
}
Abstract: Non-literal language phenomena such as idioms or metaphors are commonly studied in isolation from each other in NLP. However, often similar definitions and features are being used for different phenomena, challenging the distinction. Instead, we propose to view the detection problem as a generalized non-literal language classification problem. In this paper, we investigate multitask learning for related non-literal language phenomena. We show that in contrast to simply joining the data of multiple tasks, multi-task learning consistently improves upon four metaphor and idiom detection tasks in two languages, English and German. Comparing two state-of-theart multi-task learning architectures, we also investigate when soft parameter sharing and learned information flow can be beneficial for our related tasks. We make our adapted code publicly available.
Contact person: Erik-Lân Do Dinh, dodinh@ukp.informatik.tu-darmstadt.de
- UKP Lab: https://www.ukp.tu-darmstadt.de/
- TU Darmstadt: https://www.tu-darmstadt.de/
This project is a fork of https://github.com/sebastianruder/sluice-networks
This repository contains experimental software and is published for the sole purpose of giving additional background details on the respective publication.