BriefGPT.xyz
Jan, 2013
词嵌入的表现能力
The Expressive Power of Word Embeddings
HTML
PDF
Yanqing Chen, Bryan Perozzi, Rami Al-Rfou, Steven Skiena
TL;DR
本研究主要通过对情感极性和同义词/反义词关系的评估，比较不同嵌入方法之间的质量差异，并探讨嵌入空间中维数的多少以及每个维度的分辨率对NLP任务的影响。结果表明嵌入方法在不考虑句子结构的情况下可以捕捉到非常微妙的语义，而确定嵌入方法的质量和特性对最终结果具有重要影响。
Abstract
We seek to better understand the difference in quality of the several publicly released
embeddings
. We propose several tasks that help to distinguish the characteristics of different
embeddings
. Our evaluation sh
→