The University of Arizona at SemEval-2021 Task 10: Applying Self-training, Active Learning and Data Augmentation to Source-free Domain Adaptation

Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021), pages 458–466 Bangkok, Thailand (online), August 5–6, 2021. ©2021 Association for Computational Linguistics 458 The University of Arizona at SemEval-2021 Task 10: Applying Self-training, Active Learning and Data Augmentation to Source-free Domain Adaptation Xin Su School of Information University of Arizona xinsu@email.arizona.edu Yiyun Zhao Department of Linguistics University of Arizona yiyunzhao@email.arizona.edu Steven Bethard School of Information University of Arizona bethard@arizona.edu Abstract This paper describes our systems for negation detection and time expression recognition in SemEval 2021 Task 10, Source-Free Domain Adaptation for Semantic Processing. We show that self-training, active learning and data aug- mentation techniques can improve the general- ization ability of the model on the unlabeled target domain data without accessing source domain data. We also perform detailed ab- lation studies and error analyses for our time expression recognition systems to identify the source of the performance improvement and give constructive feedback on the temporal normalization annotation guidelines. 1 Introduction Unsupervised Domain Adaptation (UDA) is a task that generalizes knowledge acquired from a model trained on labeled data in one domain (source domain) to unlabeled data in a different domain (target domain). Conventional UDA algorithms usually require access to both source-domain and target-domain data (Ganin et al., 2016; Glorot et al., 2011; Chen et al., 2012; Louizos et al., 2016). How- ever, sharing source-domain data is often not prac- tical for clinical texts due to their highly sensitive personal information and complex data use agree- ment procedures (Laparra et al., 2020). To over- come this difficulty, Laparra et al. (2020) propose a new task of source free domain adaptation (SFDA) where only models trained on source-domain data are shared, which allows the possibility of using the information from the source-domain while re- ducing private information leakage. The biggest challenge of this task is to transfer task-related in- formation embedded in the trained models. Our team participated in both subtasks of Sem- Eval 2021 Task 10 (Laparra et al., 2021), Source Free Domain Adaptation for Semantic Processing: negation detection and time expression recognition. For both tasks, participants were given a RoBERTa model (Liu et al., 2019) fine-tuned on the source- domain, and asked to make predictions in the target- domain. The goal of the negation detection task is to pre- dict whether an event in a sentence is negated by its context. This is a binary sentence classification task. For example, given the event diarrhea and the sentence Has no diarrhea and no new lumps or masses, the goal is to predict that diarrhea is negated by its context. The goal of time expression recognition sub-task (Laparra et al., 2018) is to recognize time expres- sions in the target domain. This is a named entity recognition (NER) task. The number of entity types (inside–outside–beginning format) is 65. Entity types in this task are formally defined time entity types from the Semantically Compositional Anno- tation of Time Expressions (SCATE) (Bethard and Parker, 2016) annotation schema. For example, in 2021-02-19, 2021 will …

Content truncated to 500 words