Abstract
Functional Distributional Semantics (FDS) models the meaning of words by truth-conditional functions. This provides a natural representation for hypernymy but no guarantee that it can be learnt when FDS models are trained on a corpus. In this paper, we probe into FDS models and study the representations learnt, drawing connections between quantifications, the Distributional Inclusion Hypothesis (DIH), and the variational-autoencoding objective of FDS model training. Using synthetic data sets, we reveal that FDS models learn hypernymy on a restricted class of corpus that strictly follows the DIH. We further introduce a training objective that both enables hypernymy learning under the reverse of the DIH and improves hypernymy detection from real corpora.- Anthology ID:
- 2024.acl-long.784
- Volume:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 14625–14637
- Language:
- URL:
- https://aclanthology.org/2024.acl-long.784
- DOI:
- 10.18653/v1/2024.acl-long.784
- Cite (ACL):
- Chun Hei Lo, Wai Lam, Hong Cheng, and Guy Emerson. 2024. Distributional Inclusion Hypothesis and Quantifications: Probing for Hypernymy in Functional Distributional Semantics. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 14625–14637, Bangkok, Thailand. Association for Computational Linguistics.
- Cite (Informal):
- Distributional Inclusion Hypothesis and Quantifications: Probing for Hypernymy in Functional Distributional Semantics (Lo et al., ACL 2024)
- PDF:
- https://preview.aclanthology.org/dois-2013-emnlp/2024.acl-long.784.pdf