Reverse Operation based Data Augmentation for Solving Math Word Problems

TL;DR

This paper introduces a novel data augmentation technique for math word problems that reverses their logical structure to generate high-quality training data, improving model performance.

Contribution

It presents a new reverse operation-based data augmentation method that enhances math reasoning models by creating diverse, high-quality problems with additional knowledge points.

Findings

Improved accuracy of math problem solving models

Effective augmentation outperforms baseline methods

Enhanced reasoning capabilities in models

Abstract

Automatically solving math word problems is a critical task in the field of natural language processing. Recent models have reached their performance bottleneck and require more high-quality data for training. We propose a novel data augmentation method that reverses the mathematical logic of math word problems to produce new high-quality math problems and introduce new knowledge points that can benefit learning the mathematical reasoning logic. We apply the augmented data on two SOTA math word problem solving models and compare our results with a strong data augmentation baseline. Experimental results show the effectiveness of our approach. We release our code and data at https://github.com/yiyunya/RODA.