Style Transfer in Text: Exploration and Evaluation

By Zhenxin Fu, Xiaoye Tan, Nanyun Peng, Dongyan Zhao, Rui Yan. 2017

544 citations recorded by the source corpus.

The ability to transfer styles of texts or images, is an important measurement of the advancement of artificial intelligence (AI). However, the progress in language style transfer is lagged behind other domains, such as computer vision, mainly because of the lack of parallel data and reliable evaluation metrics. In response to the challenge of lacking parallel data, we explore learning style transfer from non-parallel data. We propose two models to achieve this goal. The key idea behind the proposed models is to learn separate content representations and style representations using adversarial networks. Considering the problem of lacking principle evaluation metrics, we propose two novel evaluation metrics that measure two aspects of style transfer: transfer strength and content preservation. We benchmark our models and the evaluation metrics on two style transfer tasks: paper-news title transfer, and positive-negative review transfer. Results show that the proposed content preservation metric is highly correlate to human judgments, and the proposed models are able to generate sentences with similar content preservation score but higher style transfer strength comparing to auto-encoder.

Subfields: Cancer Research, Computer Vision and Pattern Recognition, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cancer Research

In Computer Vision and Pattern Recognition

In Natural Language Processing

Citations in the corpus

Story Ending Prediction by Transferable BERT (2019, background): “, 2018] and text style transfer [Fu et al., 2018].”
A Recipe for Arbitrary Text Style Transfer with Large Language Models (2021, neutral): “Other Acc BLEU PPL S UPERVISED M ETHODS Cross-alignment (Shen et al., 2017) 73.4 17.6 812 Backtrans (Prabhumoye et al., 2018) 90.5 5.1 424 Multidecoder (Fu et al., 2018) 50.3 27.7 1,703 Delete-only (Li et al., 2018) 81.4 28.6 606 Delete-retrieve (Li et al., 2018) 86.2 31.1 948 Unpaired RL (Xu et al.”
CommonGen: A Constrained Text Generation Challenge for Generative Commonsense Reasoning (2020, background): “, 2019; J Kurisinkel and Chen, 2019), style (Fu et al., 2018; Luo et al., 2019b; Li et al., 2018), topics (Feng et al.”
Continual Learning for Text Classification with Information Disentanglement Based Regularization (2021, background): “…sentiment (Fu et al., 2017; John et al., 2019), content (Romanov et al., 2019; Bao et al., 2019) and syntax (Bao et al., 2019) information through supervised learning from pre-deﬁned labels (John et al., 2019) or unsupervised learning with adversarial training (Fu et al., 2017; Li et al., 2020).”
Delete, Retrieve, Generate: a Simple Approach to Sentiment and Style Transfer (2018, methodology): “Previous work has attempted to use adversarial networks (Shen et al., 2017; Fu et al., 2018) for this task, but—as we demonstrate—their outputs tend to be low-quality, as judged by human raters.”
Towards Controllable Story Generation (2018, background): “, 2017), and different styles of text (Hu et al., 2017; Ficler and Goldberg, 2017; Shen et al., 2017; Fu et al., 2017).”
Politeness Transfer: A Tag and Generate Approach (2020, background): “Current style transfer techniques (Shen et al., 2017; Hu et al., 2017; Fu et al., 2018; Yang et al., 2018; John et al., 2019) try to disentangle source style from content and then combine the content with the target style to generate the sentence in the target style.”
Style Transfer Through Back-Translation (2018, neutral): “Fu et al. (2018) explore two models for style transfer.”
Evaluation of Text Generation: A Survey (2020, methodology): “One metric that is commonly used in style transfer is the classification score obtained from a pre-trained style transfer model (Fu et al., 2018).”
CTRLsum: Towards Generic Controllable Text Summarization (2020, background): “Formally, at training time keywords are identified using source documents and ground-truth summaries while at test time they are a function of source documents and user control signal: where f denotes the automatic keyword extraction algorithm, c is the user control signal, and g control represents…”
PowerTransformer: Unsupervised Controllable Revision for Biased Language Correction (2020, background): “In unsupervised settings, a majority of work has dealt with the dearth of parallel data by using encoderdecoder setups paired with discriminators to disentangle style from content and steer generations (e.g., Shen et al., 2017; Zhang et al., 2018; Fu et al., 2018; Yang et al., 2018; Niu and Bansal, ”
Fighting Offensive Language on Social Media with Unsupervised Text Style Transfer (2018, methodology): “, 2017; Fu et al., 2018) and machine translation (Lample et al., 2018; Artetxe et al., 2018). Shen et al. (2017), Fu et al.”
Adversarial Decomposition of Text Representation (2018, background): “(2017) and the principle of “carrot and stick” (Safire, 1995), in contrast to the majority of work that promotes purely adversarial approach (Goodfellow et al., 2014; Shen et al., 2017; Fu et al., 2018; Zhu et al., 2017), we propose two additional components, the discriminator D and the motivator M ”
Controllable Paraphrase Generation with a Syntactic Exemplar (2019, methodology): “Our work is closely related to this latter category, and to the syntactically-controlled paraphrase generation of Iyyer et al. (2018) in particular, but our proposed model is different in that it simply uses a single sentence as a syntactic exemplar rather than requiring a supervised parser.”
Style Transfer as Unsupervised Machine Translation (2018, unknown): “One line of research (Hu et al. 2017; Shen et al. 2017; Fu et al. 2018) leverages the auto-encoder framework to learn an encoder and a decoder, in which the encoder constructs a latent vector by removing the style information and extracting attribute-independent content from the input sentence, and…”
Complementary Auxiliary Classifiers for Label-Conditional Text Generation (2020, methodology): “On the Yelp dataset, we compare with additional previous methods, including Cross-Align (Shen et al. 2017), MultiDecoder (Fu et al. 2018), DeleteAndRetrieve (Li et al. 2018), StyleTransformer (Dai et al. 2019), Back-Translation (Prabhumoye et al. 2018), and iVAE MI (Le Fang 2019).”
Consistency by Agreement in Zero-Shot Neural Machine Translation (2019, unknown): “For instance, it could be applied to tasks such as cross-lingual natural language inference (Conneau et al., 2018), style-transfer (Shen et al., 2017; Fu et al., 2017; Prabhumoye et al., 2018), or multi-lingual image or video captioning.”