Abstractive Summarization of Product Reviews Using Discourse Structure

We propose a novel abstractive summarization system for product reviews by taking advantage of their discourse structure. First, we apply a discourse parser to each review and obtain a discourse tree representation for every review. We then modify the discourse trees such that every leaf node only contains the aspect words. Second, we aggregate the aspect discourse trees and generate a graph. We then select a subgraph representing the most important aspects and the rhetorical relations between them using a PageRank algorithm, and transform the selected subgraph into an aspect tree. Finally, we generate a natural language summary by applying a template-based NLG framework. Quantitative and qualitative analysis of the results, based on two user studies, show that our approach significantly outperforms extractive and abstractive baselines.

[1]  Ivan Titov,et al.  A Bayesian Model for Joint Unsupervised Induction of Sentiment, Aspect and Discourse Representations , 2013, ACL.

[2]  Jackie Chi Kit Cheung,et al.  Multi-Document Summarization of Evaluative Text , 2013, EACL.

[3]  Subhabrata Mukherjee,et al.  Sentiment Aggregation using ConceptNet Ontology , 2013, IJCNLP.

[4]  Yue Lu,et al.  Rated aspect summarization of short comments , 2009, WWW '09.

[5]  Nicholas Asher,et al.  Distilling Opinion in Discourse: A Preliminary Study , 2008, COLING.

[6]  Jacob Eisenstein,et al.  Discourse Connectors for Latent Subjectivity in Sentiment Analysis , 2013, NAACL.

[7]  Lise Getoor,et al.  Supervised and Unsupervised Methods in Employing Discourse Relations for Improving Opinion Polarity Classification , 2009, EMNLP.

[8]  Ehud Reiter,et al.  Book Reviews: Building Natural Language Generation Systems , 2000, CL.

[9]  Alexander S. Yeh,et al.  More accurate tests for the statistical significance of result differences , 2000, COLING.

[10]  Bing Liu,et al.  Mining Opinion Features in Customer Reviews , 2004, AAAI.

[11]  Hugo Liu,et al.  ConceptNet — A Practical Commonsense Reasoning Tool-Kit , 2004 .

[12]  Sasha Blair-Goldensohn,et al.  Sentiment Summarization: Evaluating and Learning User Preferences , 2009, EACL.

[13]  Ivan Titov,et al.  A Joint Model of Text and Aspect Ratings for Sentiment Summarization , 2008, ACL.

[14]  Wai Lam,et al.  MEAD - A Platform for Multidocument Multilingual Text Summarization , 2004, LREC.

[15]  Regina Barzilay,et al.  Information Fusion in the Context of Multi-Document Summarization , 1999, ACL.

[16]  Bing Liu,et al.  Mining and summarizing customer reviews , 2004, KDD.

[17]  Jiawei Han,et al.  Opinosis: A Graph Based Approach to Abstractive Summarization of Highly Redundant Opinions , 2010, COLING.

[18]  Shafiq R. Joty,et al.  Combining Intra- and Multi-sentential Rhetorical Parsing for Document-level Discourse Analysis , 2013, ACL.

[19]  Jackie Chi Kit Cheung,et al.  Extractive vs. NLG-based Abstractive Summarization of Evaluative Text: The Effect of Corpus Controversiality , 2008, INLG.

[20]  ChengXiang Zhai,et al.  Comprehensive Review of Opinion Summarization , 2011 .

[21]  Robert J. Gaizauskas,et al.  A Hybrid Approach to Multi-document Summarization of Opinions in Reviews , 2014, INLG.

[22]  Bo Pang,et al.  Thumbs up? Sentiment Classification using Machine Learning Techniques , 2002, EMNLP.

[23]  D. Ariely,et al.  “Coherent Arbitrariness”: Stable Demand Curves Without Stable Preferences , 2003 .

[24]  Wenpu Xing,et al.  Weighted PageRank algorithm , 2004, Proceedings. Second Annual Conference on Communication Networks and Services Research, 2004..

[25]  Bo Pang,et al.  A Sentimental Education: Sentiment Analysis Using Subjectivity Summarization Based on Minimum Cuts , 2004, ACL.

[26]  Umeshwar Dayal,et al.  Compact explanatory opinion summarization , 2013, CIKM.

[27]  William C. Mann,et al.  Rhetorical Structure Theory: Toward a functional theory of text organization , 1988 .

[28]  Bing Liu,et al.  Opinion Feature Extraction Using Class Sequential Rules , 2006, AAAI Spring Symposium: Computational Approaches to Analyzing Weblogs.

[29]  Regina Barzilay,et al.  Multiple Aspect Ranking Using the Good Grief Algorithm , 2007, NAACL.

[30]  Dragomir R. Radev,et al.  LexRank: Graph-based Lexical Centrality as Salience in Text Summarization , 2004, J. Artif. Intell. Res..