2018/08/27 by Stefanos Angelidis, Mirella Lapata, Angelidis, Stefanos +1 · 5 citations
Computer Science · #Advanced Text Analysis Techniques #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Sentiment Analysis and Opinion Mining #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1808.08858
openalex publication_date 2018/08/27 · openalex created_date 2022/08/03 · openalex updated_date 2026/07/28
We present a neural framework for opinion summarization from online product\nreviews which is knowledge-lean and only requires light supervision (e.g., in\nthe form of product domain labels and user-provided ratings). Our method\ncombines two weakly supervised components to identify salient opinions and form\nextractive summaries from multiple reviews: an aspect extractor trained under a\nmulti-task objective, and a sentiment predictor based on multiple instance\nlearning. We introduce an opinion summarization dataset that includes a\ntraining set of product reviews from six diverse domains and human-annotated\ndevelopment and test sets with gold standard aspect annotations, salience\nlabels, and opinion summaries. Automatic evaluation shows significant\nimprovements over baselines, and a large-scale study indicates that our opinion\nsummaries are preferred by human judges according to multiple criteria.\n