Summarizing Opinions: Aspect Extraction Meets Sentiment Prediction and They Are Both Weakly Supervised

Stefanos Angelidis, Mirella Lapata
8/27/2018
cs.CLcs.AIcs.LG

Official code: stangelid/oposum (101 stars, Apache-2.0, last commit 2020-12-17).

Abstract

We present a neural framework for opinion summarization from online product reviews which is knowledge-lean and only requires light supervision (e.g., in the form of product domain labels and user-provided ratings). Our method combines two weakly supervised components to identify salient opinions and form extractive summaries from multiple reviews: an aspect extractor trained under a multi-task objective, and a sentiment predictor based on multiple instance learning. We introduce an opinion summarization dataset that includes a training set of product reviews from six diverse domains and human-annotated development and test sets with gold standard aspect annotations, salience labels, and opinion summaries. Automatic evaluation shows significant improvements over baselines, and a large-scale study indicates that our opinion summaries are preferred by human judges according to multiple criteria.

View on arXivView PDF

Code Implementations(1)

stangelid/oposumOfficial100%
Apache-2.0
10119Lua, ShellAug 21, 20185 years agoApache-2.0

Similar Papers

OpinSummEval: Revisiting Automated Evaluation for Opinion Summarization

87% match2023

SEOpinion: Summarization and Exploration Opinion of E-Commerce Websites

84% match2023

Decomposed Opinion Summarization with Verified Aspect-Aware Modules

83% match1 repo2025

Product Description and QA Assisted Self-Supervised Opinion Summarization

83% match2024

Towards Personalized Review Summarization by Modeling Historical Reviews from Customer and Product Separately

83% match2023

Opinion mining using Double Channel CNN for Recommender System

83% match2023

Cite this paper

@article{angelidis2018summarizing,
  title  = {Summarizing Opinions: Aspect Extraction Meets Sentiment Prediction and They Are Both Weakly Supervised},
  author = {Stefanos Angelidis and Mirella Lapata},
  year   = {2018},
  eprint = {1808.08858},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1808.08858}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1808.08858.svg)](https://codeofpaper.com/paper/1808.08858)

Discussion