Paraphrases do not explain word analogies
Abstract
Many types of distributional word embeddings (weakly) encode linguistic regularities as directions (the difference between "jump" and "jumped" will be in a similar direction to that of "walk" and "walked," and so on). Several attempts have been made to explain this fact. We respond to Allen and Hospedales' recent (ICML, 2019) theoretical explanation, which claims that word2vec and GloVe will encode linguistic regularities whenever a specific relation of paraphrase holds between the four words involved in the regularity. We demonstrate that the explanation does not go through: the paraphrase relations needed under this explanation do not hold empirically.
Keywords
Cite
@article{arxiv.2102.11749,
title = {Paraphrases do not explain word analogies},
author = {Louis Fournier and Ewan Dunbar},
journal= {arXiv preprint arXiv:2102.11749},
year = {2021}
}
Comments
To appear in Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers