Expansional Retrofitting for Word Vector Enrichment
This work addresses the need for improved word representations in NLP, offering incremental advancements in retrofitting techniques for tasks like word similarity and text classification.
The paper tackles the problem of enriching word vectors by proposing expansional retrofitting (extrofitting), an unsupervised method that does not require external semantic lexicons, and shows it outperforms previous methods on most word similarity tasks using GloVe vectors.
Retrofitting techniques, which inject external resources into word representations, have compensated the weakness of distributed representations in semantic and relational knowledge between words. Implicitly retrofitting word vectors by expansional technique outperforms retrofitting in word similarity tasks with word vector generalization. In this paper, we propose unsupervised extrofitting: expansional retrofitting (extrofitting) without external semantic lexicons. We also propose deep extrofitting: in-depth stacking of extrofitting and further combinations of extrofitting with retrofitting. When experimenting with GloVe, we show that our methods outperform the previous methods on most of word similarity tasks while requiring only synonyms as an external resource. Lastly, we show the effect of word vector enrichment on text classification task, as a downstream task.