Complaint focus recognition method and device based on multi-feature fusion SKIPGRAM model and generative network, and storage medium

The invention discloses a complaint focus recognition method and device based on a multi-feature fusion SKIPGRAM model and a generative network and a storage medium, and the method comprises the steps: sequentially carrying out the fusion of a feature tag and a feature tag added to each statement, obtaining a fused statement, and enabling the fused statement to form a new text statement list; the new text statement list is put into the SKIPGRAM-BILSTM-Attention complaint focus recognition network for training prediction, the probability of each statement in the new text statement list is obtained, and the two statements with the highest probability serve as dispute focuses of the two complaint parties. According to the method, the problem of insufficient features in the recognition of the complaint focus can be solved, the accumulation of legal processing predictions and the trouble of manual annotation are reduced, and the efficiency of the recognition of the complaint focus in case description is greatly im