Are Pretrained Language Models Symbolic Reasoners over Knowledge?

Nora Kassner, Benno Krojer, Hinrich Schütze


Abstract
How can pretrained language models (PLMs) learn factual knowledge from the training set? We investigate the two most important mechanisms: reasoning and memorization. Prior work has attempted to quantify the number of facts PLMs learn, but we present, using synthetic data, the first study that investigates the causal relation between facts present in training and facts learned by the PLM. For reasoning, we show that PLMs seem to learn to apply some symbolic reasoning rules correctly but struggle with others, including two-hop reasoning. Further analysis suggests that even the application of learned reasoning rules is flawed. For memorization, we identify schema conformity (facts systematically supported by other facts) and frequency as key factors for its success.
Anthology ID:
2020.conll-1.45
Volume:
Proceedings of the 24th Conference on Computational Natural Language Learning
Month:
November
Year:
2020
Address:
Online
Editors:
Raquel Fernández, Tal Linzen
Venue:
CoNLL
SIG:
SIGNLL
Publisher:
Association for Computational Linguistics
Note:
Pages:
552–564
Language:
URL:
https://aclanthology.org/2020.conll-1.45
DOI:
10.18653/v1/2020.conll-1.45
Bibkey:
Cite (ACL):
Nora Kassner, Benno Krojer, and Hinrich Schütze. 2020. Are Pretrained Language Models Symbolic Reasoners over Knowledge?. In Proceedings of the 24th Conference on Computational Natural Language Learning, pages 552–564, Online. Association for Computational Linguistics.
Cite (Informal):
Are Pretrained Language Models Symbolic Reasoners over Knowledge? (Kassner et al., CoNLL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.conll-1.45.pdf
Code
 BennoKrojer/reasoning-over-facts