Multiple-Attribute Text Style Transfer
This work addresses the challenge of fine-grained control in text style transfer for applications like content generation, though it is incremental as it builds on existing back-translation methods.
The paper tackles the problem of unsupervised text style transfer by showing that disentangled latent representations are not necessary and proposes a new model using back-translation for controlling multiple attributes like gender and sentiment, achieving better generations on new benchmarks with multiple sentences and attributes.
The dominant approach to unsupervised "style transfer" in text is based on the idea of learning a latent representation, which is independent of the attributes specifying its "style". In this paper, we show that this condition is not necessary and is not always met in practice, even with domain adversarial training that explicitly aims at learning such disentangled representations. We thus propose a new model that controls several factors of variation in textual data where this condition on disentanglement is replaced with a simpler mechanism based on back-translation. Our method allows control over multiple attributes, like gender, sentiment, product type, etc., and a more fine-grained control on the trade-off between content preservation and change of style with a pooling operator in the latent space. Our experiments demonstrate that the fully entangled model produces better generations, even when tested on new and more challenging benchmarks comprising reviews with multiple sentences and multiple attributes.