A Chinese Knowledge Extraction Model based on BERT with Fourier Transforms

The Chinese context is distinguished by robust context-dependent relationships, a multitude of named entity types, and a plethora of homotypic snear-synonyms. This underscores the pressing need for an enhancement in the feature extraction capability of the existing Chinese knowledge extraction model. This paper introduces and constructs a Fourier transform-amplitude-frequency characteristic calculation module within the BERT model, to enhance the feature extraction capability of the model. And the model has a higher F1 value on the CLUENER dataset than BERT.