Multi-Perspective Context Aggregation for Semi-supervised Cloze-style Reading Comprehension

TL;DR

This paper introduces a multi-perspective framework for cloze-style reading comprehension that aggregates diverse context information and employs a novel sampling method to enhance training data, achieving state-of-the-art results.

Contribution

The paper presents a novel multi-perspective aggregation framework combined with an automatic data augmentation technique for semi-supervised cloze reading comprehension.

Findings

Achieves new state-of-the-art performance on CLOTH dataset.

Effective multi-perspective context aggregation improves comprehension accuracy.

Sampling mechanism enhances training data quality and model robustness.

Abstract

Cloze-style reading comprehension has been a popular task for measuring the progress of natural language understanding in recent years. In this paper, we design a novel multi-perspective framework, which can be seen as the joint training of heterogeneous experts and aggregate context information from different perspectives. Each perspective is modeled by a simple aggregation module. The outputs of multiple aggregation modules are fed into a one-timestep pointer network to get the final answer. At the same time, to tackle the problem of insufficient labeled data, we propose an efficient sampling mechanism to automatically generate more training examples by matching the distribution of candidates between labeled and unlabeled data. We conduct our experiments on a recently released cloze-test dataset CLOTH (Xie et al., 2017), which consists of nearly 100k questions designed by professional teachers. Results show that our method achieves new state-of-the-art performance over previous strong baselines.