Context-Aware Answer Extraction in Question Answering

TL;DR

This paper introduces BLANC, a novel context-aware extractive question answering model that improves answer relevance by incorporating context prediction and block attention, outperforming state-of-the-art models especially when answer text occurs multiple times.

Contribution

BLANC is the first model to integrate context prediction with block attention for extractive QA, enhancing answer relevance in ambiguous passages.

Findings

BLANC outperforms state-of-the-art QA models on reading comprehension tasks.

Performance gap increases with more answer text occurrences.

BLANC excels in zero-shot settings on HotpotQA.

Abstract

Extractive QA models have shown very promising performance in predicting the correct answer to a question for a given passage. However, they sometimes result in predicting the correct answer text but in a context irrelevant to the given question. This discrepancy becomes especially important as the number of occurrences of the answer text in a passage increases. To resolve this issue, we propose \textbf{BLANC} (\textbf{BL}ock \textbf{A}ttentio\textbf{N} for \textbf{C}ontext prediction) based on two main ideas: context prediction as an auxiliary task in multi-task learning manner, and a block attention method that learns the context prediction task. With experiments on reading comprehension, we show that BLANC outperforms the state-of-the-art QA models, and the performance gap increases as the number of answer text occurrences increases. We also conduct an experiment of training the models using SQuAD and predicting the supporting facts on HotpotQA and show that BLANC outperforms all baseline models in this zero-shot setting.