EIBC: a deep learning framework for Chinese toponym recognition with multiple layers

Existing methods based on BERT are difficult to automatically identify and efficiently detect Chinese toponyms due to its irregularity and the intricate structure. To address this issue, this article introduces a novel toponym recognition model named EIBC, which is the abbreviation of ERNIE-Gram-IDCNN-BiLSTM-CRF. It consists of four parts: (1) ERNIE-Gram is selected for dynamic vector representations of toponyms and extracts toponym features; (2) the context features are dilated by IDCNN with different dilation scales; (3) BiLSTM is employed to capture bidirectional context information and to grasp a broader range of global context features, while removing the noise information through its gating mechanisms; and (4) it incorporates CRF for global optimization of toponym sequence labels, enhancing toponym recognition effectiveness. The proposed model is constructed based on a multi-layer deep learning framework by utilizing various advanced techniques to enhance the model's performance. Experimental results show that the EIBC model outperforms existing some state-of-the-art Chinese toponym recognition models.