KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph
Recently, the lexicon method has been proven to be effective for named entity recognition (NER). However, most existing lexicon-based methods cannot fully utilize common-sense knowledge in the knowledge graph. For example, the word embeddings pretrained by Word2vector or Glove lack better contextual...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-07-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/12/15/7702 |
| _version_ | 1827618534911377408 |
|---|---|
| author | Weiwei Hu Liang He Hanhan Ma Kai Wang Jingfeng Xiao |
| author_facet | Weiwei Hu Liang He Hanhan Ma Kai Wang Jingfeng Xiao |
| author_sort | Weiwei Hu |
| collection | DOAJ |
| description | Recently, the lexicon method has been proven to be effective for named entity recognition (NER). However, most existing lexicon-based methods cannot fully utilize common-sense knowledge in the knowledge graph. For example, the word embeddings pretrained by Word2vector or Glove lack better contextual semantic information usage. Hence, how to make the best of knowledge for the NER task has become a challenging and hot research topic. We propose a knowledge graph-inspired named-entity recognition (KGNER) featuring a masking and encoding method to incorporate common sense into bidirectional encoder representations from transformers (BERT). The proposed method not only preserves the original sentence semantic information but also takes advantage of the knowledge information in a more reasonable way. Subsequently, we model the temporal dependencies by taking the conditional random field (CRF) as the backend, and improve the overall performance. Experiments on four dominant datasets demonstrate that the KGNER outperforms other lexicon-based models in terms of performance. |
| first_indexed | 2024-03-09T10:09:59Z |
| format | Article |
| id | doaj.art-eab4b39dedf64e1699d2da4c118b7c19 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-09T10:09:59Z |
| publishDate | 2022-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-eab4b39dedf64e1699d2da4c118b7c192023-12-01T22:50:38ZengMDPI AGApplied Sciences2076-34172022-07-011215770210.3390/app12157702KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge GraphWeiwei Hu0Liang He1Hanhan Ma2Kai Wang3Jingfeng Xiao4College of Information Science and Engineering, Xinjiang University, Urumqi 830046, ChinaCollege of Information Science and Engineering, Xinjiang University, Urumqi 830046, ChinaCollege of Information Science and Engineering, Xinjiang University, Urumqi 830046, ChinaState Grid Xinjiang Electric Power Co., Ltd., Urumqi 830000, ChinaState Grid Xinjiang Electric Power Co., Ltd., Urumqi 830000, ChinaRecently, the lexicon method has been proven to be effective for named entity recognition (NER). However, most existing lexicon-based methods cannot fully utilize common-sense knowledge in the knowledge graph. For example, the word embeddings pretrained by Word2vector or Glove lack better contextual semantic information usage. Hence, how to make the best of knowledge for the NER task has become a challenging and hot research topic. We propose a knowledge graph-inspired named-entity recognition (KGNER) featuring a masking and encoding method to incorporate common sense into bidirectional encoder representations from transformers (BERT). The proposed method not only preserves the original sentence semantic information but also takes advantage of the knowledge information in a more reasonable way. Subsequently, we model the temporal dependencies by taking the conditional random field (CRF) as the backend, and improve the overall performance. Experiments on four dominant datasets demonstrate that the KGNER outperforms other lexicon-based models in terms of performance.https://www.mdpi.com/2076-3417/12/15/7702named-entity recognitionknowledge graphconditional random field |
| spellingShingle | Weiwei Hu Liang He Hanhan Ma Kai Wang Jingfeng Xiao KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph Applied Sciences named-entity recognition knowledge graph conditional random field |
| title | KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph |
| title_full | KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph |
| title_fullStr | KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph |
| title_full_unstemmed | KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph |
| title_short | KGNER: Improving Chinese Named Entity Recognition by BERT Infused with the Knowledge Graph |
| title_sort | kgner improving chinese named entity recognition by bert infused with the knowledge graph |
| topic | named-entity recognition knowledge graph conditional random field |
| url | https://www.mdpi.com/2076-3417/12/15/7702 |
| work_keys_str_mv | AT weiweihu kgnerimprovingchinesenamedentityrecognitionbybertinfusedwiththeknowledgegraph AT lianghe kgnerimprovingchinesenamedentityrecognitionbybertinfusedwiththeknowledgegraph AT hanhanma kgnerimprovingchinesenamedentityrecognitionbybertinfusedwiththeknowledgegraph AT kaiwang kgnerimprovingchinesenamedentityrecognitionbybertinfusedwiththeknowledgegraph AT jingfengxiao kgnerimprovingchinesenamedentityrecognitionbybertinfusedwiththeknowledgegraph |