Few-Shot Learning for Opinion Summarization

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 4119–4135, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 4119 Few-Shot Learning for Opinion Summarization Arthur Braˇzinskas1 Mirella Lapata1 Ivan Titov1,2 1ILCC, University of Edinburgh 2ILLC, University of Amsterdam abrazinskas@ed.ac.uk, {mlap,ititov}@inf.ed.ac.uk Abstract Opinion summarization is the automatic cre- ation of text reflecting subjective information expressed in multiple documents, such as user reviews of a product. The task is practically important and has attracted a lot of attention. However, due to the high cost of summary pro- duction, datasets large enough for training su- pervised models are lacking. Instead, the task has been traditionally approached with extrac- tive methods that learn to select text fragments in an unsupervised or weakly-supervised way. Recently, it has been shown that abstractive summaries, potentially more fluent and better at reflecting conflicting information, can also be produced in an unsupervised fashion. How- ever, these models, not being exposed to actual summaries, fail to capture their essential prop- erties. In this work, we show that even a hand- ful of summaries is sufficient to bootstrap gen- eration of the summary text with all expected properties, such as writing style, informative- ness, fluency, and sentiment preservation. We start by training a conditional Transformer lan- guage model to generate a new product review given other available reviews of the product. The model is also conditioned on review prop- erties that are directly related to summaries; the properties are derived from reviews with no manual effort. In the second stage, we fine-tune a plug-in module that learns to pre- dict property values on a handful of summaries. This lets us switch the generator to the summa- rization mode. We show on Amazon and Yelp datasets that our approach substantially outper- forms previous extractive and abstractive meth- ods in automatic and human evaluation. 1 Introduction Summarization of user opinions expressed in on- line resources, such as blogs, reviews, social media, or internet forums, has drawn much attention due to its potential for various information access appli- cations, such as creating digests, search, and report Gold These shoes run true to size, do a good job supporting the arch of the foot and are well-suited for exercise. They’re good looking, comfortable, and the sole feels soft and cushioned. Overall they are a nice, light-weight pair of shoes and come in a variety of stylish colors. Ours These running shoes are great! They fit true to size and are very comfortable to run around in. They are light weight and have great support. They run a little on the narrow side, so make sure to order a half size larger than normal. Reviews perfect fit for me ... supply the support that I need ... are flexible and comfort- able ... || ... It is very comfortable ... I enjoy wearing them running ... || ... running shoes ... felt great right out of the box ... They run true to size ... || ... my feet and feel like …

Content truncated to 500 words