Unsupervised Opinion Summarization with Noising and Denoising

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1934–1945 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 1934 Unsupervised Opinion Summarization with Noising and Denoising Reinald Kim Amplayo and Mirella Lapata Institute for Language, Cognition and Computation School of Informatics, University of Edinburgh reinald.kim@ed.ac.uk, mlap@inf.ed.ac.uk Abstract The supervised training of high-capacity mod- els on large datasets containing hundreds of thousands of document-summary pairs is criti- cal to the recent success of deep learning tech- niques for abstractive summarization. Unfortu- nately, in most domains (other than news) such training data is not available and cannot be eas- ily sourced. In this paper we enable the use of supervised learning for the setting where there are only documents available (e.g., prod- uct or business reviews) without ground truth summaries. We create a synthetic dataset from a corpus of user reviews by sampling a re- view, pretending it is a summary, and gener- ating noisy versions thereof which we treat as pseudo-review input. We introduce several linguistically motivated noise generation func- tions and a summarization model which learns to denoise the input and generate the original review. At test time, the model accepts gen- uine reviews and generates a summary contain- ing salient opinions, treating those that do not reach consensus as noise. Extensive automatic and human evaluation shows that our model brings substantial improvements over both ab- stractive and extractive baselines. 1 Introduction The proliferation of massive numbers of online product, service, and merchant reviews has pro- vided strong impetus to develop systems that per- form opinion mining automatically (Pang and Lee, 2008). The vast majority of previous work (Hu and Liu, 2006) breaks down the problem of opin- ion aggregation and summarization into three inter- related tasks involving aspect extraction (Mukher- jee and Liu, 2012), sentiment identification (Pang et al., 2002; Pang and Lee, 2004), and summary creation based on extractive (Radev et al., 2000; Lu et al., 2009) or abstractive methods (Ganesan et al., 2010; Carenini et al., 2013; Gerani et al., 2014; Di Fabbrizio et al., 2014). Although po- tentially more challenging, abstractive approaches seem more appropriate for generating informative and concise summaries, e.g., by performing var- ious rewrite operations (e.g., deletion of words or phrases and insertion of new ones) which go beyond simply copying and rearranging passages from the original opinions. Abstractive summarization has enjoyed renewed interest in recent years thanks to the availability of large-scale datasets (Sandhaus, 2008; Hermann et al., 2015; Grusky et al., 2018; Liu et al., 2018; Fabbri et al., 2019) which have driven the devel- opment of neural architectures for summarizing single and multiple documents. Several approaches (See et al., 2017; Celikyilmaz et al., 2018; Paulus et al., 2018; Gehrmann et al., 2018; Liu et al., 2018; Perez-Beltrachini et al., 2019; Liu and La- pata, 2019; Wang and Ling, 2016) have shown promising results with sequence-to-sequence mod- els that encode one or several source documents and then decode the learned representations into an abstractive summary. …

Content truncated to 500 words