How Can BERT Help Lexical Semantics Tasks?
This addresses the problem of adapting contextual embeddings for word-level tasks in NLP, though it appears incremental as it builds on existing BERT technology.
The paper tackles the limitation of BERT's dynamic embeddings for lexical semantics tasks by using them to train static embeddings, achieving state-of-the-art results on seven datasets.
Contextualized embeddings such as BERT can serve as strong input representations to NLP tasks, outperforming their static embeddings counterparts such as skip-gram, CBOW and GloVe. However, such embeddings are dynamic, calculated according to a sentence-level context, which limits their use in lexical semantics tasks. We address this issue by making use of dynamic embeddings as word representations in training static embeddings, thereby leveraging their strong representation power for disambiguating context information. Results show that this method leads to improvements over traditional static embeddings on a range of lexical semantics tasks, obtaining the best reported results on seven datasets.