CDUMA: An Adaptive Approach for Mitigating Confounder for MCQA
Shilong Li, Chenji Lu, Ge Bai, Ying Liu, Xiyan Liu, Zhang Zhang, Ruifang Liu
Abstract
Multiple-choice question answering (MCQA) requires the model to select the correct answer from a set of candidate options when given a passage and a question. Previous research has achieved promising results with the assistance of Pre-trained Language Models(PrLMs). However, it has been observed that these models heavily rely on text similarity for inference. In this study, we approach the MCQA task from a causal perspective, treating text similarity as a confounder and establishing a structural causal model(SCM) for the MCQA task. To mitigate the impact of this confounder, we propose a novel Causal Dual Multi-head Co-Attention (CDUMA) model. CDUMA reduces the influence of text similarity between questions and options on model inference and enhances the model’s generalization ability during routine testing. Experimental results on two widely used datasets demonstrate the effectiveness of our approach.
BibTeX
@inproceedings{icassp2024_cdumaanadaptivea,
title = {CDUMA: An Adaptive Approach for Mitigating Confounder for MCQA},
author = {Shilong Li and Chenji Lu and Ge Bai and Ying Liu and Xiyan Liu and Zhang Zhang and Ruifang Liu},
booktitle = {ICASSP 2024},
year = {2024}
}