Enhancing the QA Model through a Multi-domain Debiasing Framework

TL;DR

This paper presents a multi-domain debiasing framework that improves the robustness of QA models against biases and adversarial challenges, leading to significant performance gains.

Contribution

It introduces a novel multi-domain debiasing approach combining knowledge distillation and domain expansion for QA models.

Findings

Up to 2.6 percentage point improvements in EM and F1 scores.

Enhanced robustness of QA models in adversarial settings.

Effective bias mitigation across multiple domains.

Abstract

Question-answering (QA) models have advanced significantly in machine reading comprehension but often exhibit biases that hinder their performance, particularly with complex queries in adversarial conditions. This study evaluates the ELECTRA-small model on the Stanford Question Answering Dataset (SQuAD) v1.1 and adversarial datasets AddSent and AddOneSent. By identifying errors related to lexical bias, numerical reasoning, and entity recognition, we develop a multi-domain debiasing framework incorporating knowledge distillation, debiasing techniques, and domain expansion. Our results demonstrate up to 2.6 percentage point improvements in Exact Match (EM) and F1 scores across all test sets, with gains in adversarial contexts. These findings highlight the potential of targeted bias mitigation strategies to enhance the robustness and reliability of natural language understanding systems.