Leveraging Integrated Learning for Open-Domain Chinese Named Entity Recognition
Citations Over Time
Abstract
Named entity recognition (NER) is a fundamental technique in natural language processing that provides preconditions for tasks, such as natural language question reasoning, text matching, and semantic text similarity. Compared to English, the challenge of Chinese NER lies in the noise impact caused by the complex meanings, diverse structures, and ambiguous semantic boundaries of the Chinese language itself. At the same time, compared with specific domains, open-domain entity types are more complex and changeable, and the number of entities is considerably larger. Thus, the task of Chinese NER is more difficult. However, existing open-domain NER methods have low recognition rates. Therefore, this paper proposes a method based on the bidirectional long short-term memory conditional random field (BiLSTM-CRF) model, which leverages integrated learning to improve the efficiency of Chinese NER. Compared with single models, including CRF, BiLSTM-CRF, and gated recurrent unit-CRF, the proposed method can significantly improve the accuracy of open-domain Chinese NER.
Related Papers
- → Hybrid semi-Markov CRF for Neural Sequence Labeling(2018)56 cited
- Conditional Random Fields vs. Hidden Markov Models in a biomedical Named Entity Recognition task(2007)
- → CRF-based active learning for Chinese named entity recognition(2009)20 cited
- → Joint Self-Attention and Multi-Embeddings for Chinese Named Entity Recognition(2020)12 cited
- → Hybrid semi-Markov CRF for Neural Sequence Labeling(2018)3 cited