Semantics Altering Modifications for Evaluating Comprehension in Machine Reading

TL;DR

This paper evaluates whether current machine reading comprehension models can handle Semantics Altering Modifications, revealing that despite high performance, models often fail to process these linguistically challenging alterations.

Contribution

The paper introduces a method for generating challenge sets with semantic modifications and a novel evaluation approach to assess MRC models' understanding of altered semantics.

Findings

Models perform poorly on SAM-enriched data despite high overall accuracy.

Evaluation methodology effectively isolates semantic understanding from domain effects.

Large-scale study covers 12 architectures and 4 datasets, highlighting widespread limitations.

Abstract

Advances in NLP have yielded impressive results for the task of machine reading comprehension (MRC), with approaches having been reported to achieve performance comparable to that of humans. In this paper, we investigate whether state-of-the-art MRC models are able to correctly process Semantics Altering Modifications (SAM): linguistically-motivated phenomena that alter the semantics of a sentence while preserving most of its lexical surface form. We present a method to automatically generate and align challenge sets featuring original and altered examples. We further propose a novel evaluation methodology to correctly assess the capability of MRC systems to process these examples independent of the data they were optimised on, by discounting for effects introduced by domain shift. In a large-scale empirical study, we apply the methodology in order to evaluate extractive MRC models with regard to their capability to correctly process SAM-enriched data. We comprehensively cover 12 different state-of-the-art neural architecture configurations and four training datasets and find that -- despite their well-known remarkable performance -- optimised models consistently struggle to correctly process semantically altered data.