Style Transfer from Non-Parallel Text by Cross-Alignment

TL;DR

This paper introduces a cross-alignment method for style transfer in non-parallel text, effectively separating content from style across various tasks like sentiment change, cipher decipherment, and word order recovery.

Contribution

It proposes a novel approach leveraging refined alignment of latent representations to perform style transfer without parallel data, applicable to multiple NLP tasks.

Findings

Effective style transfer demonstrated on sentiment modification

Successful decipherment of word substitution ciphers

Recovery of original word order in text

Abstract

This paper focuses on style transfer on the basis of non-parallel text. This is an instance of a broad family of problems including machine translation, decipherment, and sentiment modification. The key challenge is to separate the content from other aspects such as style. We assume a shared latent content distribution across different text corpora, and propose a method that leverages refined alignment of latent representations to perform style transfer. The transferred sentences from one style should match example sentences from the other style as a population. We demonstrate the effectiveness of this cross-alignment method on three tasks: sentiment modification, decipherment of word substitution ciphers, and recovery of word order.