Towards Controllable and Personalized Review Generation

TL;DR

This paper introduces RevGAN, a new model for generating controllable, personalized reviews that outperform existing models in quality, coherence, and realism, using novel autoencoder and discriminator components.

Contribution

The paper presents RevGAN, a novel review generation model with three innovative components enabling controllability and personalization, outperforming prior models.

Findings

RevGAN generates reviews indistinguishable from real ones.

RevGAN outperforms state-of-the-art models in quality and coherence.

Generated reviews follow natural linguistic laws.

Abstract

In this paper, we propose a novel model RevGAN that automatically generates controllable and personalized user reviews based on the arbitrarily given sentimental and stylistic information. RevGAN utilizes the combination of three novel components, including self-attentive recursive autoencoders, conditional discriminators, and personalized decoders. We test its performance on the several real-world datasets, where our model significantly outperforms state-of-the-art generation models in terms of sentence quality, coherence, personalization and human evaluations. We also empirically show that the generated reviews could not be easily distinguished from the organically produced reviews and that they follow the same statistical linguistics laws.