Research on Domain Specific Chinese Named Entity Recognition Based on RTBC Algorithm

In the task of Chinese named entity recognition, how to enhance the recognition ability of the model for the boundary between characters and words and how to process the common polysemy of words is a hot issue that many scholars are working on it. In this paper, we propose a Chinese entity recognition model incorporating language model, RoBERTa-WWM-TextCNN-BiGRU-CRF model, which uses RoBERTa model pretrained on large-scale corpus to dynamically generate word vector sequence according to its input context, and then uses BiGRU and TextCNN combined model to further extract sentence features and capture word boundary information, and finally input the sequences of feature vectors into the final prediction results are achieved by inputting some constraints into the CRF model. Experiments were performed on the resume dataset and a customized dataset foreign affair, and the precision, recall, and F1 values were all improved compared to current mainstream named entity recognition models.