Abstract: Chinese passive sentences is an important linguistic phenomenon. In this paper, we use the BIO combined with indexing annotation method to annotate the passive structures in passive sentences at a fine-grained level. We propose a CRF sequence labeling model for passive structure parsing in Chinese based on the BERT-wwm-ext pre-training model and the biaffine attention mechanism. The proposed model exhibits a significant improvement in F₁ value of 97.31%, which can effectively enhance the performance of passive sentence parsing in CAMR by post-processing.