Study on Named Entity Recognition Based on Graph Convolutional Neural Network

In the field of electric power operation & inspection, a substantial amount of data are usually generated. However, these data are often unstructured and their expression is relatively complex. Nevertheless, these data contain great value. To address the problem of extracting valuable structured information from these unstructured power texts, A model specialized in identifying named entities in Chinese within the domain of electric power operation & inspection is proposed in this paper. It can effectively models the grammatical relationships between sentences and impoves how textual features are represented by adding grammatical information. At the same time, it uses bidirectional gated recurrent control units to capture temporal relationships and the interaction between multiple variables, thereby obtaining more effective local features. The experimental results show that the F1 score of the proposed model in the paper reaches 0.88646, and its performance is better than that of the traditional named entity recognition methods, which proves the effectiveness of the model.