Journal of the Korea Society of Computer and Information (한국컴퓨터정보학회논문지)

Korean Society of Computer Information (한국컴퓨터정보학회)
권호 표지
DOI QR코드

DOI QR Code

DeNERT: Named Entity Recognition Model using DQN and BERT

  • Received : 2020.02.13
  • Accepted : 2020.04.09
  • Published : 2020.04.29
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, we propose a new structured entity recognition DeNERT model. Recently, the field of natural language processing has been actively researched using pre-trained language representation models with a large amount of corpus. In particular, the named entity recognition, which is one of the fields of natural language processing, uses a supervised learning method, which requires a large amount of training dataset and computation. Reinforcement learning is a method that learns through trial and error experience without initial data and is closer to the process of human learning than other machine learning methodologies and is not much applied to the field of natural language processing yet. It is often used in simulation environments such as Atari games and AlphaGo. BERT is a general-purpose language model developed by Google that is pre-trained on large corpus and computational quantities. Recently, it is a language model that shows high performance in the field of natural language processing research and shows high accuracy in many downstream tasks of natural language processing. In this paper, we propose a new named entity recognition DeNERT model using two deep learning models, DQN and BERT. The proposed model is trained by creating a learning environment of reinforcement learning model based on language expression which is the advantage of the general language model. The DeNERT model trained in this way is a faster inference time and higher performance model with a small amount of training dataset. Also, we validate the performance of our model's named entity recognition performance through experiments.

본 논문에서는 새로운 구조의 개체명 인식 DeNERT 모델을 제안한다. 최근 자연어처리 분야는 방대한 양의 말뭉치로 사전 학습된 언어 표현 모델을 활용하는 연구가 활발하다. 특히 자연어처리 분야 중 하나인 개체명인식은 대부분 지도학습 방식을 사용하는데, 충분히 많은 양의 학습 데이터 세트와 학습 연산량이 필요하다는 단점이 있다. 강화학습은 초기 데이터 없이 시행착오 경험을 통해 학습하는 방식으로 다른 기계학습 방법론보다 조금 더 사람이 학습하는 과정에 가까운 알고리즘으로 아직 자연어처리 분야에는 많이 적용되지 않은 분야이다. 아타리 게임이나 알파고 등 시뮬레이션 가능한 게임 환경에서 많이 사용된다. BERT는 대량의 말뭉치와 연산량으로 학습된 구글에서 개발한 범용 언어 모델이다. 최근 자연어 처리 연구 분야에서 높은 성능을 보이고 있는 언어 모델이며 많은 자연어처리 하위분야에서도 높은 정확도를 나타낸다. 본 논문에서는 이러한 DQN, BERT 두가지 딥러닝 모델을 이용한 새로운 구조의 개체명 인식 DeNERT 모델을 제안한다. 제안하는 모델은 범용 언어 모델의 장점인 언어 표현력을 기반으로 강화학습 모델의 학습 환경을 만드는 방법으로 학습된다. 이러한 방식으로 학습된 DeNERT 모델은 적은 양의 학습 데이터세트로 더욱 빠른 추론시간과 높은 성능을 갖는 모델이다. 마지막으로 제안하는 모델의 개체명 인식 성능평가를 위해 실험을 통해서 검증한다.

Keywords

References

  1. T. Mikolov and I. Sutskever, "Distributed representations of words and phrases and their compositionality," Advances in neural information processing systems, pp. 3111-3119, Oct. 2013. DOI: 1310.4546
  2. G. Lample and M. Ballesteros, "Neural architectures for named entity recognition," Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp.260-270, Jun. 2016. DOI: 10.18653/v1/N16-1030
  3. ME. Peters and M. Neumann, "Deep contextualized word representations," Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pp.2227-2237, Jun. 2018. DOI: 10.18653/v1/N18-1202
  4. A. Radford and K. Narasimhan, "Improving language understanding by generative pre-training," URL https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf
  5. J. Devlin and MW. Chang, "Bert: Pre-training of deep bidirectional transformers for language understanding," Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pp.4171-4186, Jun. 2019. DOI: 10.18653/v1/N19-1423
  6. A. Vaswani and N. Shazeer, "Attention is all you need," Advances in neural information processing systems, Dec. 2017. https://arxiv.org/abs/1706.03762
  7. CJCH. Watkins and P.Dayan, "Q-learning," Machine learning 8.3-4. pp. 279-292, May. 1992. DOI: 10.1007/BF00992698
  8. V. Mnih and K. Kavukcuoglu, "Playing atari with deep reinforcement learning," arXiv preprint arXiv:1312.5602, Dec. 2013. https://arxiv.org/abs/1312.5602
  9. V. Mnih and K. Kavukcuoglu, "Human-level control through deep reinforcement learning," Nature 518.7540, pp.529-533, Feb. 2015. DOI: 10.1038/nature14236
  10. M. Fang and Y. Li, "Learning how to active learn: A deep reinforcement learning approach," Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 595-605. Sep. 2017. DOI: 10.18653/v1/D17-1063
  11. Y. Yang and W. Chen, "Distantly supervised ner with partial annotation learning and reinforcement learning," Proceedings of the 27th International Conference on Computational Linguistics, pp. 2159-2169, Aug. 2018. https://www.aclweb.org/anthology/C18-1183.pdf
  12. Z. Huang and W. Xu, "Bidirectional LSTM-CRF models for sequence tagging," Pro-ceedings of the 21st International Conference on AsianLanguage Processing, Aug. 2015. https://arxiv.org/abs/1508.01991
  13. Y. Wu and M. Schuster, "Google's neural machine translation system: Bridging the gap between human and machine translation," arXiv preprint arXiv:1609.08144, Oct. 2016. https://arxiv.org/abs/1609.08144
  14. CD. Manning and M. Surdeanu, "The Stanford CoreNLP natural language processing toolkit," Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations, pp.55-60, Jun. 2014. DOI: 10.3115/v1/P14-5010
  15. EF. Sang and F. De. Meulder, "Introduction to the CoNLL-2003 shared task: Language-independent named entity recognition," Proceedings of the Seventh Conference on Natural Language Learning at HLT-NAACL, pp.142-147, Jun. 2003. https://arxiv.org/abs/cs/0306050
  16. M. Abadi and A. Agarwal, "Tensorflow: Large-scale machine learning on heterogeneous distributed systems," arXiv preprint arXiv:1603.04467, Mar. 2016. https://arxiv.org/abs/1603.04467
  17. A. Akbik and T. Bergmann, "Pooled contextualized embeddings for named entity recognition," Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics, Volume 1, pp. 724-728, Jun. 2019. DOI: 10.18653/v1/N19-1078
  18. V. Mnih and AP. Badia, "Asynchronous methods for deep reinforcement learning," International conference on machine learning, Feb. 2016. https://arxiv.org/abs/1602.01783
  19. RS. Sutton and DA. McAllester, "Policy gradient methods for reinforcement learning with function approximation," Advances in neural information processing systems, pp. 1057-1063, Jun. 2000. https://arxiv.org/abs/1706.06643