“…As the long short-term memory network (LSTM) method has excellent sequence modeling ability and automatically detects word-level features, it has been widely used for name entity recognition (NER) tasks, and most of the new methods of name entity recognition are based on LSTM models ( Li et al, 2022 ; Chen et al, 2022 ; Rajput et al, 2022 ; Karim et al, 2022 ). However, as the inherent network architecture of the recurrent neural network, LSTM and recurrent neural network (RNN) variant models ignore the text structure and the semantic connection of the context, the word vector represented by the ordinary method is only the static word vector before the word-level information and does not contain the semantic information of the context ( Ravenscroft, Goetze & Hain, 2022 ; Cui, Li & Xu, 2022 ; Li et al, 2021 ; Lang et al, 2021 ).…”