RankAug: Augmented data ranking for text classification

TL;DR

RankAug introduces a text-ranking filtering method for synthetic data in NLP, significantly improving classification accuracy for under-represented classes by selecting diverse, meaning-preserving augmented texts.

Contribution

The paper presents RankAug, a novel data filtering approach that enhances synthetic data quality for text classification tasks, addressing a gap in evaluation methods for generated data.

Findings

Up to 35% improvement in classification accuracy for under-represented classes.

Effective filtering techniques enhance the utility of augmented data.

Demonstrated across multiple datasets and NLP tasks.

Abstract

Research on data generation and augmentation has been focused majorly on enhancing generation models, leaving a notable gap in the exploration and refinement of methods for evaluating synthetic data. There are several text similarity metrics within the context of generated data filtering which can impact the performance of specific Natural Language Understanding (NLU) tasks, specifically focusing on intent and sentiment classification. In this study, we propose RankAug, a text-ranking approach that detects and filters out the top augmented texts in terms of being most similar in meaning with lexical and syntactical diversity. Through experiments conducted on multiple datasets, we demonstrate that the judicious selection of filtering techniques can yield a substantial improvement of up to 35% in classification accuracy for under-represented classes.