Decomposing Textual Information For Style Transfer

Abstract

This paper focuses on latent representations that could effectively decompose different aspects of textual information. Using a framework of style transfer for texts, we propose several empirical methods to assess information decomposition quality. We validate these methods with several state-of-the-art textual style transfer methods. Higher quality of information decomposition corresponds to higher performance in terms of bilingual evaluation understudy (BLEU) between output and human-written reformulations.

arXiv commentarXiv admin note: substantial text overlap with arXiv:1908.06809

References in corpus (9)

• Style Transfer from Non-Parallel Text by Cross-Alignment
• Style Transfer as Unsupervised Machine Translation
• On Unifying Deep Generative Models
• Unpaired Sentiment-to-Sentiment Translation: A Cycled Reinforcement Learning Approach
• Multiple-Attribute Text Style Transfer
• Evaluating prose style transfer with the Bible
• Structured Content Preservation for Unsupervised Text Style Transfer
• Generating lyrics with variational autoencoder and multi-modal artist embeddings
• What is wrong with style transfer for texts?

Cited by in corpus (4)

• Style-transfer and Paraphrase: Looking for a Sensible Semantic Similarity Metric
• From Theories on Styles to their Transfer in Text: Bridging the Gap with a Hierarchical Survey
• Are "Undocumented Workers" the Same as "Illegal Aliens"? Disentangling Denotation and Connotation in Vector Spaces
• Paranoid Transformer: Reading Narrative of Madness as Computational Approach to Creativity