Chinese Named Entity Recognition Based on Multi-feature fusion

Chinese Named Entity Recognition(NER) is a critical technology for processing Chinese text information, its core goal is to identify and extract named entities with specific semantic meaning from unstructured text. Although the technology of Chinese named entity recognition is constantly developing, there is still a problem that it is not possible to make full use of the unique information of Chinese characters such as glyphes and ‘pinyin’ to deeply mine the rich semantics contained in Chinese. To address the above challenges, this paper proposes a Chinese named entity recognition mode, MFF-NER, based on multi-feature fusion. The model uses convolutional neural network and multi-head self-attention mechanism to capture the grapheme features and ‘pinyin’ features hidden in Chinese characters, integrates multi-dimensional information, and then uses bidirectional long short-term memory neural network to learn long-distance context representation. Finally, conditional random fields are integrated to model constraints in sentence sequence labels, thereby enhancing the model’s deep semantic mining capabilities and improving overall performance. Experimental results prove that the proposed model can effectively improve the performance of Chinese named entity recognition task.