TransSum: Translating Aspect and Sentiment Embeddings for Self-Supervised Opinion Summarization

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 729–742 August 1–6, 2021. ©2021 Association for Computational Linguistics 729 TransSum: Translating Aspect and Sentiment Embeddings for Self-Supervised Opinion Summarization Ke Wang, Xiaojun Wan Wangxuan Institute of Computer Technology, Peking University The MOE Key Laboratory of Computational Linguistics, Peking University {wangke17,wanxiaojun}@pku.edu.cn Abstract In this paper, we propose a novel self- supervised opinion summarization framework TransSum, which models opinion sum- maries as translations operating on the low-dimensional aspect and sentiment em- bedding spaces. Specifically, we propose two contrastive objectives to learn the cru- cial aspect and sentiment embeddings of reviews, by taking advantage of the intra- and inter-group invariances that have not been considered in previous studies. Furthermore, these embeddings can be used to reduce opinion redundancy and construct highly relevant reviews-summary pairs to train a supervised multi-input opinion summarization model. Experimental results on three different domains show that TransSum outperforms several strong baselines in generating informa- tive, relevant and low-redundant summaries, unveiling the effectiveness of our approach. 1 Introduction Opinion summarization, which focuses on auto- matically generating summaries that reflect salient opinion information expressed in a group of docu- ments (e.g., user reviews of a product in Figure 1), has been receiving great attention due to its use- fulness and effectiveness for displaying massive opinion texts (Ku et al., 2006; Cheung et al., 2009; Chu and Liu, 2019). For example, a representative review summary of a product can not only replace large amounts of reviews for potential customers to read, but also provide more explanations than a simple overall sentiment rating, such as “ What is the biggest complaint on the iPod ‘screen’ ?”. However, compared with supervised summariza- tion in the domain of news articles, the annotated training data for opinion summarization is expen- sive to acquire. Due to the lack of gold-standard summaries for training, most existing works fo- cus on unsupervised opinion summarization and Positive review Negative review a A sA … … Groups of reviews Yes, HP dvd’s are better quality. … I have had a ton a pro- blems with these discs. … The box was too big and allow the spindle to bo- unce around … Entity 1 Their Christmas deco- rations rival that of coach house … Cheap drinks, great happy hour … Cheap drinks, awesome bar staff, stiff pours … Entity 2 Figure 1: The proposed TransSum targets at learn- ing corresponding aspect and sentiment embeddings for reviews (green arrows) through contrastive learning based on the aspect and sentiment invariances (blue ar- rows). These embeddings are used to construct reviews- summary pairs of high relevance (red arrows), so as to train a supervised multi-input opinion summarization model. Best view in color. treat it as a normal multi-document summarization task. They either struggle to reduce the opinion redundancy efficiently or output summaries lack- ing relevance to input reviews. Particularly, many previous studies focus on extractive approaches (Paul et al., 2010; Fabbrizio et al., 2014; Rossiello et al., 2017; Narayan et al., 2019), which …

Content truncated to 500 words