CLSep 26, 2019

Decomposing Textual Information For Style Transfer

arXiv:1909.12928v11004 citations
Originality Synthesis-oriented
AI Analysis

This work addresses style transfer in text generation, but it is incremental as it builds on existing methods without introducing a new paradigm.

The paper tackled the problem of decomposing textual information for style transfer by proposing empirical methods to assess decomposition quality, and found that higher decomposition quality correlates with improved BLEU scores, such as a 5% increase in performance.

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.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes