English

Unsupervised Distillation of Syntactic Information from Contextualized Word Representations

Computation and Language 2021-03-15 v2 Machine Learning

Abstract

Contextualized word representations, such as ELMo and BERT, were shown to perform well on various semantic and syntactic tasks. In this work, we tackle the task of unsupervised disentanglement between semantics and structure in neural language representations: we aim to learn a transformation of the contextualized vectors, that discards the lexical semantics, but keeps the structural information. To this end, we automatically generate groups of sentences which are structurally similar but semantically different, and use metric-learning approach to learn a transformation that emphasizes the structural component that is encoded in the vectors. We demonstrate that our transformation clusters vectors in space by structural properties, rather than by lexical semantics. Finally, we demonstrate the utility of our distilled representations by showing that they outperform the original contextualized representations in a few-shot parsing setting.

Keywords

Cite

@article{arxiv.2010.05265,
  title  = {Unsupervised Distillation of Syntactic Information from Contextualized Word Representations},
  author = {Shauli Ravfogel and Yanai Elazar and Jacob Goldberger and Yoav Goldberg},
  journal= {arXiv preprint arXiv:2010.05265},
  year   = {2021}
}

Comments

Accepted in BlackboxNLP@EMNLP2020

R2 v1 2026-06-23T19:15:11.141Z