Improving Event Duration Prediction via Time-aware Pre-training

Zonglin Yang, Xinya Du, Alexander Rush, Claire Cardie


Abstract
End-to-end models in NLP rarely encode external world knowledge about length of time. We introduce two effective models for duration prediction, which incorporate external knowledge by reading temporal-related news sentences (time-aware pre-training). Specifically, one model predicts the range/unit where the duration value falls in (R-PRED); and the other predicts the exact duration value (E-PRED). Our best model – E-PRED, substantially outperforms previous work, and captures duration information more accurately than R-PRED. We also demonstrate our models are capable of duration prediction in the unsupervised setting, outperforming the baselines.
Anthology ID:
2020.findings-emnlp.302
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2020
Month:
November
Year:
2020
Address:
Online
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3370–3378
Language:
URL:
https://aclanthology.org/2020.findings-emnlp.302
DOI:
10.18653/v1/2020.findings-emnlp.302
Bibkey:
Cite (ACL):
Zonglin Yang, Xinya Du, Alexander Rush, and Claire Cardie. 2020. Improving Event Duration Prediction via Time-aware Pre-training. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 3370–3378, Online. Association for Computational Linguistics.
Cite (Informal):
Improving Event Duration Prediction via Time-aware Pre-training (Yang et al., Findings 2020)
Copy Citation:
PDF:
https://preview.aclanthology.org/starsem-semeval-split/2020.findings-emnlp.302.pdf
Data
MC-TACO