Few-Shot Learning for Opinion Summarization

By Arthur Bravzinskas, Mirella Lapata, Ivan Titov. 2020

89 citations recorded by the source corpus.

Opinion summarization is the automatic creation 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 production, datasets large enough for training supervised models are lacking. Instead, the task has been traditionally approached with extractive 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. However, these models, not being exposed to actual summaries, fail to capture their essential properties. In this work, we show that even a handful of summaries is sufficient to bootstrap generation of the summary text with all expected properties, such as writing style, informativeness, fluency, and sentiment preservation. We start by training a conditional Transformer language model to generate a new product review given other available reviews of the product. The model is also conditioned on review properties 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 predict property values on a handful of summaries. This lets us switch the generator to the summarization mode. We show on Amazon and Yelp datasets that our approach substantially outperforms previous extractive and abstractive methods in automatic and human evaluation.

Subfields: Artificial Intelligence, Summarization, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yu Bai
2. Yang Gao
3. Xiuzhen Zhang
4. Wenyi Tay
5. Quan Z. Sheng
6. Hu Wang
7. Sarvnaz Karimi
8. Mingyu Guo

Citations in the corpus

• LongForm: Effective Instruction Tuning with Reverse Instructions (2023, neutral): “7B, despite both having similar number of parameters, which aligns with findings from Touvron et al. (2023a).”
• Multi-document Summarization via Deep Learning Techniques: A Survey (2020, background): “Transformer HT [98] Pros: Good performance; good parallel computing; can capture cross-document and in-document relations Cons: Time-consuming; problems with position encoding MGSum [74] FewSum [17] GraphSum [94] Bart-Long [126] WikiSum [97]”
• PSP: Pre-trained Soft Prompts for Few-Shot Abstractive Summarization (2022, background): “Bražinskas et al. (2020) introduce plug-in networks to reproduce characteristics of the target dataset with only a small set of labeled examples.”
• Measuring Similarity of Opinion-bearing Sentences (2021, methodology): “…of reviews, NLP research has focused on methods to make sense of this vast data resource, including applications for opinion summarization (Suhara et al., 2020; Bražinskas et al., 2020b; Mukherjee et al., 2020; Chu and Liu, 2019; Angelidis and Lapata, 2018) and opinion search (Poddar et al., 2017).”