Neural Bi-Lexicalized PCFG Induction

Songlin Yang, Yanpeng Zhao, Kewei Tu


Abstract
Neural lexicalized PCFGs (L-PCFGs) have been shown effective in grammar induction. However, to reduce computational complexity, they make a strong independence assumption on the generation of the child word and thus bilexical dependencies are ignored. In this paper, we propose an approach to parameterize L-PCFGs without making implausible independence assumptions. Our approach directly models bilexical dependencies and meanwhile reduces both learning and representation complexities of L-PCFGs. Experimental results on the English WSJ dataset confirm the effectiveness of our approach in improving both running speed and unsupervised parsing performance.
Anthology ID:
2021.acl-long.209
Volume:
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
Month:
August
Year:
2021
Address:
Online
Venues:
ACL | IJCNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2688–2699
Language:
URL:
https://aclanthology.org/2021.acl-long.209
DOI:
10.18653/v1/2021.acl-long.209
Bibkey:
Copy Citation:
PDF:
https://aclanthology.org/2021.acl-long.209.pdf
Code
 sustcsonglin/TN-PCFG
Data
Penn Treebank