Filling method based on explicit and implicit fusion auxiliary information

The invention discloses a filling method of fusion auxiliary information based on explicit and implicit expressions, which is characterized in that missing enterprise performance information is accurately predicted by fusing implicit preference assessment items of enterprise performance and utilizing a modern deep learning noise reduction encoder model. According to the method, display numerical value information of enterprise performance is utilized, implicit preference assessment items of the enterprise performance are fused to construct implicit data, higher-dimension enterprise performance feature vectors are extracted on the basis of a noise reduction encoder, missing enterprise performance data are filled in combination with first N pieces of most similar enterprise performance numerical value information, and the enterprise performance evaluation accuracy is improved. The defect of fixed value filling is well overcome, all missing enterprise performance information is considered to be consistent based