A Simple Recipe towards Reducing Hallucination in Neural Surface Realisation

Feng Nie, Jin-Ge Yao, Jinpeng Wang, Rong Pan, Chin-Yew Lin


Abstract
Recent neural language generation systems often hallucinate contents (i.e., producing irrelevant or contradicted facts), especially when trained on loosely corresponding pairs of the input structure and text. To mitigate this issue, we propose to integrate a language understanding module for data refinement with self-training iterations to effectively induce strong equivalence between the input data and the paired text. Experiments on the E2E challenge dataset show that our proposed framework can reduce more than 50% relative unaligned noise from the original data-text pairs. A vanilla sequence-to-sequence neural NLG model trained on the refined data has improved on content correctness compared with the current state-of-the-art ensemble generator.
Anthology ID:
P19-1256
Volume:
Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics
Month:
July
Year:
2019
Address:
Florence, Italy
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2673–2679
Language:
URL:
https://www.aclweb.org/anthology/P19-1256
DOI:
10.18653/v1/P19-1256
Bib Export formats:
BibTeX MODS XML EndNote
PDF:
http://aclanthology.lst.uni-saarland.de/P19-1256.pdf
Video:
 https://vimeo.com/384728744