Style Transfer Through Multilingual and Feedback-Based Back-Translation

TL;DR

This paper introduces two extensions to existing style transfer models that enhance meaning preservation and transfer accuracy by leveraging multilingual and feedback-based back-translation techniques.

Contribution

It proposes novel extensions to style transfer models using multilingual and feedback-based back-translation to better preserve meaning and improve transfer accuracy.

Findings

Extensions improve meaning preservation in style transfer

Extensions enhance style transfer accuracy

Models grounded better in semantic content

Abstract

Style transfer is the task of transferring an attribute of a sentence (e.g., formality) while maintaining its semantic content. The key challenge in style transfer is to strike a balance between the competing goals, one to preserve meaning and the other to improve the style transfer accuracy. Prior research has identified that the task of meaning preservation is generally harder to attain and evaluate. This paper proposes two extensions of the state-of-the-art style transfer models aiming at improving the meaning preservation in style transfer. Our evaluation shows that these extensions help to ground meaning better while improving the transfer accuracy.