On Learning Text Style Transfer with Direct Rewards
This work addresses the challenge of text style transfer for natural language processing applications, offering an incremental improvement over existing methods.
The paper tackles the problem of text style transfer without parallel corpora by optimizing reward functions that assess style and content preservation, resulting in significant gains in both automatic and human evaluations over strong baselines.
In most cases, the lack of parallel corpora makes it impossible to directly train supervised models for the text style transfer task. In this paper, we explore training algorithms that instead optimize reward functions that explicitly consider different aspects of the style-transferred outputs. In particular, we leverage semantic similarity metrics originally used for fine-tuning neural machine translation models to explicitly assess the preservation of content between system outputs and input texts. We also investigate the potential weaknesses of the existing automatic metrics and propose efficient strategies of using these metrics for training. The experimental results show that our model provides significant gains in both automatic and human evaluation over strong baselines, indicating the effectiveness of our proposed methods and training strategies.