Improving Distributional Similarity with Lessons Learned from Word Embeddings
Published
2015-05-04
Omer Levy
,
Yoav Goldberg
,
Ido Dagan
Omer Levy
Bar-Ilan University
Yoav Goldberg
Bar-Ilan University
Ido Dagan
Bar-Ilan University
Abstract
Recent trends suggest that neural-network-inspired word embedding models outperform traditional count-based distributional models on word similarity and analogy detection tasks. We reveal that much of the performance gains of word embeddings are due to certain system design choices and hyperparameter optimizations, rather than the embedding algorithms themselves. Furthermore, we show that these modifications can be transferred to traditional distributional models, yielding similar gains. In contrast to prior reports, we observe mostly local or insignificant performance differences between the methods, with no global advantage to any single approach over the others.
PDF (presented at ACL 2015)