Wan2Vec: Embeddings Learned on Word Association Norms

Tracking #: 1963-3176

This paper is currently under review
Gemma Bel-Enguix
Helena Gomez-Adorno
Jorge Reyes-Magaña
Gerardo Sierra

Responsible editor: 
Guest Editors Knowledge Graphs 2018

Submission type: 
Full Paper
Word embeddings are powerful for many tasks in natural language processing. In this work, we learn word embeddings using weighted graphs from Word Association Norms with the node2vec algorithm. The computational resources used by this technique are reasonable and affordable, which allows us to obtain good quality word embeddings even from small corpus. We evaluate our word vectors in two word similarity benchmarks, the WordSim-353, MC30, MTurk-287, MEN-TR-3k, SimLex-999, MTurk-771 and RG-65, achieving better results than those obtained with word2vec, GloVe, and FastText, trained on huge corpus.
Full PDF Version: 
Under Review