Contrastive Learning on LLM Back Generation Treebank for Cross-domain Constituency Parsing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Peiming, Zhang, Meishan, Li, Jianling, Zhang, Min, Zhang, Yue
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915307297177600
author Guo, Peiming
Zhang, Meishan
Li, Jianling
Zhang, Min
Zhang, Yue
author_facet Guo, Peiming
Zhang, Meishan
Li, Jianling
Zhang, Min
Zhang, Yue
contents Cross-domain constituency parsing is still an unsolved challenge in computational linguistics since the available multi-domain constituency treebank is limited. We investigate automatic treebank generation by large language models (LLMs) in this paper. The performance of LLMs on constituency parsing is poor, therefore we propose a novel treebank generation method, LLM back generation, which is similar to the reverse process of constituency parsing. LLM back generation takes the incomplete cross-domain constituency tree with only domain keyword leaf nodes as input and fills the missing words to generate the cross-domain constituency treebank. Besides, we also introduce a span-level contrastive learning pre-training strategy to make full use of the LLM back generation treebank for cross-domain constituency parsing. We verify the effectiveness of our LLM back generation treebank coupled with contrastive learning pre-training on five target domains of MCTB. Experimental results show that our approach achieves state-of-the-art performance on average results compared with various baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2505_20976
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Contrastive Learning on LLM Back Generation Treebank for Cross-domain Constituency Parsing
Guo, Peiming
Zhang, Meishan
Li, Jianling
Zhang, Min
Zhang, Yue
Computation and Language
Cross-domain constituency parsing is still an unsolved challenge in computational linguistics since the available multi-domain constituency treebank is limited. We investigate automatic treebank generation by large language models (LLMs) in this paper. The performance of LLMs on constituency parsing is poor, therefore we propose a novel treebank generation method, LLM back generation, which is similar to the reverse process of constituency parsing. LLM back generation takes the incomplete cross-domain constituency tree with only domain keyword leaf nodes as input and fills the missing words to generate the cross-domain constituency treebank. Besides, we also introduce a span-level contrastive learning pre-training strategy to make full use of the LLM back generation treebank for cross-domain constituency parsing. We verify the effectiveness of our LLM back generation treebank coupled with contrastive learning pre-training on five target domains of MCTB. Experimental results show that our approach achieves state-of-the-art performance on average results compared with various baselines.
title Contrastive Learning on LLM Back Generation Treebank for Cross-domain Constituency Parsing
topic Computation and Language
url https://arxiv.org/abs/2505.20976