Relation Extraction Based on BERT and BGRU in the Chinese Music Scene

Due to the lack of music knowledge extraction tools and public datasets, there are relatively few studies on music entity relation extraction. The Bidirectional Encoder Representations from Transformer (BERT) model considers contextual information and provides more dynamic, richer vector representations of generated words. Bidirectional Gated Recurrent Units (BGRU) has longdistance learning capability, which can obtain rich contextual semantic feature. Therefore, this paper proposes a deep learning model based on BERT and BGRU combined with an attention mechanism for Chinese music relation extraction. Experimental results show that the proposed method obtains good results when extracting the relation between musical entities.