SemGloVe: Semantic Co-occurrences for GloVe from BERT
This work provides an incremental improvement for researchers and practitioners using GloVe word embeddings by enhancing their semantic representation.
This paper proposes SemGloVe, a method to improve GloVe word embeddings by distilling semantic co-occurrences from BERT. It addresses the limitations of GloVe's local context window by extracting word pairs and defining co-occurrence weights based on BERT's masked language model or multi-head attention, leading to improved performance over GloVe on word similarity datasets and four external tasks.
GloVe learns word embeddings by leveraging statistical information from word co-occurrence matrices. However, word pairs in the matrices are extracted from a predefined local context window, which might lead to limited word pairs and potentially semantic irrelevant word pairs. In this paper, we propose SemGloVe, which distills semantic co-occurrences from BERT into static GloVe word embeddings. Particularly, we propose two models to extract co-occurrence statistics based on either the masked language model or the multi-head attention weights of BERT. Our methods can extract word pairs without limiting by the local window assumption and can define the co-occurrence weights by directly considering the semantic distance between word pairs. Experiments on several word similarity datasets and four external tasks show that SemGloVe can outperform GloVe.