Multi-span Style Extraction for Generative Reading Comprehension

TL;DR

This paper introduces a multi-span extraction framework for generative reading comprehension, improving answer quality by combining the strengths of extractive and generative models.

Contribution

It proposes a novel multi-span extraction approach that enhances generative MRC performance, addressing issues of incomplete and redundant answers.

Findings

Improved answer quality with better syntax and semantics.

Alleviates the gap between generative and extractive models.

Demonstrates effectiveness through thorough experiments.

Abstract

Generative machine reading comprehension (MRC) requires a model to generate well-formed answers. For this type of MRC, answer generation method is crucial to the model performance. However, generative models, which are supposed to be the right model for the task, in generally perform poorly. At the same time, single-span extraction models have been proven effective for extractive MRC, where the answer is constrained to a single span in the passage. Nevertheless, they generally suffer from generating incomplete answers or introducing redundant words when applied to the generative MRC. Thus, we extend the single-span extraction method to multi-span, proposing a new framework which enables generative MRC to be smoothly solved as multi-span extraction. Thorough experiments demonstrate that this novel approach can alleviate the dilemma between generative models and single-span models and produce answers with better-formed syntax and semantics.