We proposed a novel approach to train word embeddings to capture antonyms in our NAACL/HLT-2015 paper . Word embeddings have shown to capture synonyms and analogies. Such word embeddings, however, cannot capture antonyms since they depend on the distributional hypothesis. Our approach utilizes supervised synonym and antonym information from thesauri, as well as distributional information from large-scale unlabelled text data. The evaluation results on the GRE antonym question task show that our
model outperforms the state-of-the-art systems and it can answer the antonym questions at the F-score of 89%.
 Masataka Ono, Makoto Miwa, Yutaka Sasaki, Word Embedding-based Antonym Detection using Thesauri and Distributional Information, NAACL/HLT-2015, 2015.
Antonym Detection Web Demonstration
(Caution: This page can be used only for the demo purpose. Accessing more than 100 times per day is not allowed.)