CASIA at SemEval-2022 Task 11: Chinese Named Entity Recognition for Complex and Ambiguous Entities
Jia Fu | Zhen Gan | Zhucong Li | Sirui Li | Dianbo Sui | Yubo Chen | Kang Liu | Jun Zhao
Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022)
This paper describes our approach to develop a complex named entity recognition system in SemEval 2022 Task 11: MultiCoNER Multilingual Complex Named Entity Recognition,Track 9 - Chinese. In this task, we need to identify the entity boundaries and categorylabels for the six identified categories of CW,LOC, PER, GRP, CORP, and PORD.The task focuses on detecting semantically ambiguous and complex entities in short and low-context settings. We constructed a hybrid system based on Roberta-large model with three training mechanisms and a series of data gugmentation.Three training mechanisms include adversarial training, Child-Tuning training, and continued pre-training. The core idea of the hybrid system is to improve the performance of the model in complex environments by introducing more domain knowledge through data augmentation and continuing pre-training domain adaptation of the model. Our proposed method in this paper achieves a macro-F1 of 0.797 on the final test set, ranking second.
CASIA@SMM4H’22: A Uniform Health Information Mining System for Multilingual Social Media Texts
Jia Fu | Sirui Li | Hui Ming Yuan | Zhucong Li | Zhen Gan | Yubo Chen | Kang Liu | Jun Zhao | Shengping Liu
Proceedings of The Seventh Workshop on Social Media Mining for Health Applications, Workshop & Shared Task
This paper presents a description of our system in SMM4H-2022, where we participated in task 1a,task 4, and task 6 to task 10. There are three main challenges in SMM4H-2022, namely the domain shift problem, the prediction bias due to category imbalance, and the noise in informal text. In this paper, we propose a unified framework for the classification and named entity recognition tasks to solve the challenges, and it can be applied to both English and Spanish scenarios. The results of our system are higher than the median F1-scores for 7 tasks and significantly exceed the F1-scores for 6 tasks. The experimental results demonstrate the effectiveness of our system.