Bidder accurate extraction method and system based on HMM and BI-LSTM model

The invention discloses a bidder accurate extraction method and system based on an HMM (Hidden Markov Model) and a BI-LSTM (Bit-Long Short Term Memory) model. Belongs to the field of data acquisition, and relates to a bidder accurate extraction method based on HMM and BI-LSTM models, which comprises the following steps: carrying out corpus preliminary extraction on corpus data of an original document, and constructing a corpus data set; performing data role labeling on the corpus data set according to the type of the bidder to obtain a role labeling data set; and adopting HMM and BI-LSTM models to carry out automatic labeling on the role labeling data set, then judging a bidder and outputting the bidder. According to the method, bidders are divided into different components for classification, the classification effect is effectively improved, a BI-LSTM and HMM multi-model fusion algorithm model is provided, and the mathematical model is fully utilized to be actually applied to specific bidder identification