论文 ID ATH-D99ZJN · 一区论文 · 计算机科学

Enriching Word Vectors with Subword Information CC-BY

这是一篇从 Transactions of the Association for Computational Linguistics 导入的开放获取论文,用于长期保存;它并非本站投稿,因此不设版本历史与时间戳历史。

  • 原出版日期:2017年12月1日 UTC
  • 权利状态:CC-BY(未记录届满日期)
  • 加入本站:2026年10月3日 UTC
  • Piotr Bojanowski Meta (United States)
  • Édouard Grave Meta (United States)
  • Armand Joulin Meta (United States)
  • Tomáš Mikolov Meta (United States)

提交者 AthenXiv UW2JTS24Y · 发表时间 2026年10月3日 UTC · 316 浏览量 · 18 下载量 · 2.2 MB · English

由以下人员代作者上传 AthenXiv (UW2JTS24Y)

在线阅读 下载 PDF (2.2 MB)

摘要

Continuous word representations, trained on large unlabeled corpora are useful for many natural language processing tasks. Popular models that learn such representations ignore the morphology of words, by assigning a distinct vector to each word. This is a limitation, especially for languages with large vocabularies and many rare words. In this paper, we propose a new approach based on the skipgram model, where each word is represented as a bag of character n-grams. A vector representation is associated to each character n-gram; words being represented as the sum of these representations. Our method is fast, allowing to train models on large corpora quickly and allows us to compute word representations for words that did not appear in the training data. We evaluate our word representations on nine different languages, both on word similarity and analogy tasks. By comparing to recently proposed morphological word representations, we show that our vectors achieve state-of-the-art performance on these tasks.

关键词: Topic Modeling, Artificial Intelligence, Computer Science, Physical Sciences

DOI: 10.1162/tacl_a_00051

许可协议: CC-BY

在线阅读

引用本文

BibTeX RIS 纯文本 JSON

Piotr Bojanowski, Édouard Grave, Armand Joulin, Tomáš Mikolov (2026). Enriching Word Vectors with Subword Information. AthenXiv. https://athenxiv.com/paper/ATH-D99ZJN