Does BERT Know that the IS-A Relation Is Transitive?

Ruixi Lin, Hwee Tou Ng


Abstract
The success of a natural language processing (NLP) system on a task does not amount to fully understanding the complexity of the task, typified by many deep learning models. One such question is: can a black-box model make logically consistent predictions for transitive relations? Recent studies suggest that pre-trained BERT can capture lexico-semantic clues from words in the context. However, to what extent BERT captures the transitive nature of some lexical relations is unclear. From a probing perspective, we examine WordNet word senses and the IS-A relation, which is a transitive relation. That is, for senses A, B, and C, A is-a B and B is-a C entail A is-a C. We aim to quantify how much BERT agrees with the transitive property of IS-A relations, via a minimalist probing setting. Our investigation reveals that BERT’s predictions do not fully obey the transitivity property of the IS-A relation.
Anthology ID:
2022.acl-short.11
Volume:
Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
Month:
May
Year:
2022
Address:
Dublin, Ireland
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
94–99
Language:
URL:
https://aclanthology.org/2022.acl-short.11
DOI:
10.18653/v1/2022.acl-short.11
Bibkey:
Cite (ACL):
Ruixi Lin and Hwee Tou Ng. 2022. Does BERT Know that the IS-A Relation Is Transitive?. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 94–99, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
Does BERT Know that the IS-A Relation Is Transitive? (Lin & Ng, ACL 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingestion-script-update/2022.acl-short.11.pdf
Software:
 2022.acl-short.11.software.zip
Code
 nusnlp/probe-bert-transitivity