Towards Fine-grained Text Sentiment Transfer

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2013–2022 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 2013 Towards Fine-grained Text Sentiment Transfer Fuli Luo1, Peng Li2, Pengcheng Yang1, Jie Zhou2, Yutong Tan3, Baobao Chang1,4, Zhifang Sui1,4, Xu Sun1 1Key Lab of Computational Linguistics, Peking University 2Pattern Recognition Center, WeChat AI, Tencent Inc, China 3Computer Science and Technology, Beijing Normal University 4Peng Cheng Laboratory, China luofuli@pku.edu.cn, patrickpli@tencent.com, yang pc@pku.edu.cn, withtomzhou@tencent.com, tanyt@mail.bnu.edu.cn, {chbb,szf,xusun}@pku.edu.cn Abstract In this paper, we focus on the task of fine- grained text sentiment transfer (FGST). This task aims to revise an input sequence to satisfy a given sentiment intensity, while preserving the original semantic content. Different from conventional sentiment transfer task that only reverses the sentiment polarity (posi- tive/negative) of text, the FTST task requires more nuanced and fine-grained control of sentiment. To remedy this, we propose a novel Seq2SentiSeq model. Specifically, the numeric sentiment intensity value is incorporated into the decoder via a Gaussian kernel layer to finely control the sentiment intensity of the output. Moreover, to tackle the problem of lacking parallel data, we propose a cycle reinforcement learning algorithm to guide the model training. In this framework, the elaborately designed rewards can balance both sentiment transformation and content preservation, while not requiring any ground truth output. Experimental results show that our approach can outperform existing methods by a large margin in both automatic evaluation and human evaluation. Our code and data, including outputs of atall baselines and our model are available https://github.com/luofuli/ Fine-grained-Sentiment-Transfer. 1 1 Introduction Text sentiment transfer aims to rephrase the in- put to satisfy a given sentiment label (value) while preserving its original semantic content. It facil- itates various NLP applications, such as automat- ically converting the attitude of review and fight- ing against offensive language in social media (dos Santos et al., 2018). Previous work (Shen et al., 2017; Li et al., 2018; Luo et al., 2019) on text sentiment transfer mainly focuses on the coarse-grained level: the reversal of 1Joint work between WeChat AI and Peking University. Input Sentence Tasty food and wonderful service. Target Sentiment Output Sentence 0.1 Horrible food and terrible service! 0.3 Plain food, slow service. 0.5 Food and service need improvement. 0.7 Good food and service. 0.9 Amazing food and perfect service!! Tar Senti 0 0 0 0 0 Input S Figure 1: An example of the input and output of the fine-grained text sentiment transfer task. The output reviews describe the same content (e.g. food/service) as the input while expressing different sentiment intensity. positive and negative sentiment polarity. They are confined to scenarios where there are two discrete sentiment labels. To achieve more nuanced and precise sentiment control of text generation, we turn to fine-grained text sentiment transfer (FTST) which revises a sequence to satisfy a given senti- ment intensity2, while keeping the semantic con- tent unchanged. Taking Figure 1 as an example, given the same input and five sentiment intensity values ranging from …

Content truncated to 500 words