Counter-fitting Word Vectors to Linguistic Constraints
This work addresses the need for more accurate semantic representations in natural language processing, particularly for tasks like dialogue systems, though it is incremental as it builds on existing pre-trained vectors.
The authors tackled the problem of improving word vectors' semantic similarity judgments by injecting antonymy and synonymy constraints, achieving new state-of-the-art performance on the SimLex-999 dataset and robust improvements in dialogue state tracking across domains.
In this work, we present a novel counter-fitting method which injects antonymy and synonymy constraints into vector space representations in order to improve the vectors' capability for judging semantic similarity. Applying this method to publicly available pre-trained word vectors leads to a new state of the art performance on the SimLex-999 dataset. We also show how the method can be used to tailor the word vector space for the downstream task of dialogue state tracking, resulting in robust improvements across different dialogue domains.