SubmissionNumber#=%=#15
FinalPaperTitle#=%=#DKE-Research at SemEval-2024 Task 2: Incorporating Data Augmentation with Generative Models and Biomedical Knowledge to Enhance Inference Robustness
ShortPaperTitle#=%=#
NumberOfPages#=%=#7
CopyrightSigned#=%=#Yuqi Wang
JobTitle#==#
Organization#==#
Abstract#==#Safe and reliable natural language inference is critical for extracting insights from clinical trial reports but poses challenges due to biases in large pre-trained language models. This paper presents a novel data augmentation technique to improve model robustness for biomedical natural language inference in clinical trials. By generating synthetic examples through semantic perturbations and domain-specific vocabulary replacement and adding a new task for numerical and quantitative reasoning, we introduce greater diversity and reduce shortcut learning. Our approach, combined with multi-task learning and the DeBERTa architecture, achieved significant performance gains on the NLI4CT 2024 benchmark compared to the original language models. Ablation studies validate the contribution of each augmentation method in improving robustness. Our best-performing model ranked 12th in terms of faithfulness and 8th in terms of consistency, respectively, out of the 32 participants.
Author{1}{Firstname}#=%=#Yuqi
Author{1}{Lastname}#=%=#Wang
Author{1}{Username}#=%=#yuqiw99
Author{1}{Email}#=%=#yuqi.wang17@student.xjtlu.edu.cn
Author{1}{Affiliation}#=%=#Xi'an Jiaotong Liverpool University
Author{2}{Firstname}#=%=#Zeqiang
Author{2}{Lastname}#=%=#Wang
Author{2}{Username}#=%=#zeqiangwang
Author{2}{Email}#=%=#zeqiangwang.medicalai@outlook.com
Author{2}{Affiliation}#=%=#University of Surrey
Author{3}{Firstname}#=%=#Wei
Author{3}{Lastname}#=%=#Wang
Author{3}{Email}#=%=#wei.wang03@xjtlu.edu.cn
Author{3}{Affiliation}#=%=#Xi'an Jiaotong Liverpool University
Author{4}{Firstname}#=%=#Qi
Author{4}{Lastname}#=%=#Chen
Author{4}{Email}#=%=#qi.chen02@xjtlu.edu.cn
Author{4}{Affiliation}#=%=#Xi'an Jiaotong Liverpool University
Author{5}{Firstname}#=%=#Kaizhu
Author{5}{Lastname}#=%=#Huang
Author{5}{Username}#=%=#khuang_cn
Author{5}{Email}#=%=#kaizhu.huang@dukekunshan.edu.cn
Author{5}{Affiliation}#=%=#Duke Kunshan University
Author{6}{Firstname}#=%=#Anh
Author{6}{Lastname}#=%=#Nguyen
Author{6}{Email}#=%=#anh.nguyen@liverpool.ac.uk
Author{6}{Affiliation}#=%=#University of Liverpool
Author{7}{Firstname}#=%=#Suparna
Author{7}{Lastname}#=%=#De
Author{7}{Username}#=%=#supdey
Author{7}{Email}#=%=#s.de@surrey.ac.uk
Author{7}{Affiliation}#=%=#University of Surrey

==========
èéáğö