Machine Reading Comprehension using Case-based Reasoning
This work addresses the need for reliable and debuggable QA systems by offering an interpretable approach, though it is incremental as it builds on existing case-based reasoning ideas.
The paper tackles answer extraction in machine reading comprehension by proposing an interpretable case-based reasoning method (CBR-MRC) that retrieves similar cases and predicts answers based on semantic similarities, achieving high accuracy with improvements of 11.5 and 8.4 EM on NaturalQuestions and NewsQA datasets.
We present an accurate and interpretable method for answer extraction in machine reading comprehension that is reminiscent of case-based reasoning (CBR) from classical AI. Our method (CBR-MRC) builds upon the hypothesis that contextualized answers to similar questions share semantic similarities with each other. Given a test question, CBR-MRC first retrieves a set of similar cases from a nonparametric memory and then predicts an answer by selecting the span in the test context that is most similar to the contextualized representations of answers in the retrieved cases. The semi-parametric nature of our approach allows it to attribute a prediction to the specific set of evidence cases, making it a desirable choice for building reliable and debuggable QA systems. We show that CBR-MRC provides high accuracy comparable with large reader models and outperforms baselines by 11.5 and 8.4 EM on NaturalQuestions and NewsQA, respectively. Further, we demonstrate the ability of CBR-MRC in identifying not just the correct answer tokens but also the span with the most relevant supporting evidence. Lastly, we observe that contexts for certain question types show higher lexical diversity than others and find that CBR-MRC is robust to these variations while performance using fully-parametric methods drops.