Subword-augmented Embedding for Cloze Reading Comprehension

TL;DR

This paper introduces a subword-augmented embedding method for cloze reading comprehension, improving word representations by using subword units to handle rare words and enhance model performance.

Contribution

It proposes a novel subword-based embedding augmentation strategy and empirically demonstrates its effectiveness over character-based methods in reading comprehension tasks.

Findings

Significant performance improvement over baselines on multiple datasets

Effective handling of rare words with subword augmentation

Enhanced generalization ability of the reading model

Abstract

Representation learning is the foundation of machine reading comprehension. In state-of-the-art models, deep learning methods broadly use word and character level representations. However, character is not naturally the minimal linguistic unit. In addition, with a simple concatenation of character and word embedding, previous models actually give suboptimal solution. In this paper, we propose to use subword rather than character for word embedding enhancement. We also empirically explore different augmentation strategies on subword-augmented embedding to enhance the cloze-style reading comprehension model reader. In detail, we present a reader that uses subword-level representation to augment word embedding with a short list to handle rare words effectively. A thorough examination is conducted to evaluate the comprehensive performance and generalization ability of the proposed reader. Experimental results show that the proposed approach helps the reader significantly outperform the state-of-the-art baselines on various public datasets.