{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,11]],"date-time":"2026-09-11T16:43:37Z","timestamp":1789145017289,"version":"build-2803163510"},"reference-count":56,"publisher":"MIT Press","license":[{"start":{"date-parts":[[2021,9,10]],"date-time":"2021-09-10T00:00:00Z","timestamp":1631232000000},"content-version":"vor","delay-in-days":252,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2021,9,8]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>This paper presents a novel unsupervised abstractive summarization method for opinionated texts. While the basic variational autoencoder-based models assume a unimodal Gaussian prior for the latent code of sentences, we alternate it with a recursive Gaussian mixture, where each mixture component corresponds to the latent code of a topic sentence and is mixed by a tree-structured topic distribution. By decoding each Gaussian component, we generate sentences with tree-structured topic guidance, where the root sentence conveys generic content, and the leaf sentences describe specific topics. Experimental results demonstrate that the generated topic sentences are appropriate as a summary of opinionated texts, which are more informative and cover more input contents than those generated by the recent unsupervised summarization model (Bra\u017einskas et al., 2020). Furthermore, we demonstrate that the variance of latent Gaussians represents the granularity of sentences, analogous to Gaussian word embedding (Vilnis and McCallum, 2015).<\/jats:p>","DOI":"10.1162\/tacl_a_00406","type":"journal-article","created":{"date-parts":[[2021,9,20]],"date-time":"2021-09-20T15:24:03Z","timestamp":1632151443000},"page":"945-961","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":8,"title":["Unsupervised Abstractive Opinion Summarization by Generating Sentences with Tree-Structured Topic Guidance"],"prefix":"10.1162","volume":"9","author":[{"given":"Masaru","family":"Isonuma","sequence":"first","affiliation":[{"name":"The University of Tokyo, Japan. isonuma@ipr-ctr.t.u-tokyo.ac.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Junichiro","family":"Mori","sequence":"additional","affiliation":[{"name":"The University of Tokyo, Japan"},{"name":"RIKEN, Japan. mori@mi.u-tokyo.ac.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Danushka","family":"Bollegala","sequence":"additional","affiliation":[{"name":"University of Liverpool, United Kingdom. danushka@liverpool.ac.uk"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ichiro","family":"Sakata","sequence":"additional","affiliation":[{"name":"The University of Tokyo, Japan. isakata@ipr-ctr.t.u-tokyo.ac.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"281","published-online":{"date-parts":[[2021,9,8]]},"reference":[{"key":"2021091013024592300_bib1","article-title":"Tree-structured decoding with doubly-recurrent neural networks","volume-title":"Proceedings of the 5th International Conference on Learning Representations","author":"Alvarez-Melis","year":"2017"},{"key":"2021091013024592300_bib2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.175","article-title":"Unsupervised opinion summarization with content planning","author":"Amplayo","year":"2020","journal-title":"CoRR"},{"key":"2021091013024592300_bib3","doi-asserted-by":"crossref","first-page":"1934","DOI":"10.18653\/v1\/2020.acl-main.175","article-title":"Unsupervised opinion summarization with noising and denoising","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Amplayo","year":"2020"},{"key":"2021091013024592300_bib4","doi-asserted-by":"publisher","first-page":"277","DOI":"10.1162\/tacl_a_00366","article-title":"Extractive opinion summarization in quantized transformer spaces","volume":"9","author":"Angelidis","year":"2021","journal-title":"Transactions of the Association for Computational Linguistics"},{"key":"2021091013024592300_bib5","doi-asserted-by":"publisher","first-page":"3675","DOI":"10.18653\/v1\/D18-1403","article-title":"Summarizing opinions: Aspect extraction meets sentiment prediction and they are both weakly supervised","volume-title":"Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing","author":"Angelidis","year":"2018"},{"issue":"1","key":"2021091013024592300_bib6","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1162\/coli.2008.34.1.1","article-title":"Modeling local coherence: An entity-based approach","volume":"34","author":"Barzilay","year":"2008","journal-title":"Computational Linguistics"},{"key":"2021091013024592300_bib7","first-page":"673","article-title":"Seq\u22273: Differentiable sequence-to-sequence- to-sequence autoencoder for unsupervised abstractive sentence compression","volume-title":"Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","author":"Baziotis","year":"2019"},{"key":"2021091013024592300_bib8","doi-asserted-by":"publisher","first-page":"10","DOI":"10.18653\/v1\/K16-1002","article-title":"Generating sentences from a continuous space","volume-title":"Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning","author":"Bowman","year":"2016"},{"key":"2021091013024592300_bib9","doi-asserted-by":"publisher","first-page":"5151","DOI":"10.18653\/v1\/2020.acl-main.461","article-title":"Unsupervised opinion summarization as copycat-review generation","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Bra\u017einskas","year":"2020"},{"key":"2021091013024592300_bib10","doi-asserted-by":"publisher","first-page":"335","DOI":"10.1145\/290941.291025","article-title":"The use of MMR, diversity-based reranking for reordering documents and producing summaries","volume-title":"Proceedings of the 21st International ACM SIGIR Conference on Research and Development in Information Retrieval","author":"Carbonell","year":"1998"},{"issue":"4","key":"2021091013024592300_bib11","doi-asserted-by":"publisher","first-page":"545","DOI":"10.1111\/j.1467-8640.2012.00417.x","article-title":"Multi-document summarization of evaluative text","volume":"29","author":"Carenini","year":"2013","journal-title":"Computational Intelligence"},{"key":"2021091013024592300_bib12","first-page":"815","article-title":"A hybrid hierarchical model for multi-document summarization","volume-title":"Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics","author":"Celikyilmaz","year":"2010"},{"key":"2021091013024592300_bib13","first-page":"491","article-title":"Discovery of topically coherent sentences for extractive summarization","volume-title":"Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies","author":"Celikyilmaz","year":"2011"},{"key":"2021091013024592300_bib14","first-page":"1","article-title":"Semeval-2017 task 1: Semantic textual similarity multilingual and crosslingual focused evaluation","volume-title":"Proceedings of the 11th International Workshop on Semantic Evaluation","author":"Cer","year":"2017"},{"key":"2021091013024592300_bib15","first-page":"1223","article-title":"Meansum: A neural model for unsupervised multi-document abstractive summarization","volume-title":"Proceedings of the 36th International Conference on Machine Learning","author":"Chu","year":"2019"},{"key":"2021091013024592300_bib16","article-title":"Empirical evaluation of gated recurrent neural networks on sequence modeling","volume-title":"Proceedings of NIPS 2014 Workshop on Deep Learning","author":"Chung","year":"2014"},{"key":"2021091013024592300_bib17","first-page":"4171","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","volume-title":"Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","author":"Devlin","year":"2019"},{"key":"2021091013024592300_bib18","first-page":"365","article-title":"Lexpagerank: Prestige in multi-document text summarization","volume-title":"Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing","author":"Erkan","year":"2004"},{"key":"2021091013024592300_bib19","doi-asserted-by":"crossref","first-page":"1074","DOI":"10.18653\/v1\/P19-1102","article-title":"Multi-news: A large-scale multi-document summarization dataset and abstractive hierarchical model","volume-title":"Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics","author":"Fabbri","year":"2019"},{"key":"2021091013024592300_bib20","doi-asserted-by":"publisher","first-page":"413","DOI":"10.18653\/v1\/K18-1040","article-title":"Unsupervised sentence compression using denoising auto-encoders","volume-title":"Proceedings of the 22nd Conference on Computational Natural Language Learning","author":"Fevry","year":"2018"},{"key":"2021091013024592300_bib21","first-page":"340","article-title":"Opinosis: A graph-based approach to abstractive summarization of highly redundant opinions","volume-title":"Proceedings of the 23rd International Conference on Computational Linguistics","author":"Ganesan","year":"2010"},{"key":"2021091013024592300_bib22","doi-asserted-by":"publisher","first-page":"1602","DOI":"10.3115\/v1\/D14-1168","article-title":"Abstractive summarization of product reviews using discourse structure","volume-title":"Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing","author":"Gerani","year":"2014"},{"key":"2021091013024592300_bib23","doi-asserted-by":"publisher","first-page":"10","DOI":"10.3115\/1611638.1611640","article-title":"A scalable global model for summarization","volume-title":"Proceedings of the Workshop on Integer Linear Programming for Natural Language Processing","author":"Gillick","year":"2009"},{"key":"2021091013024592300_bib24","first-page":"17","article-title":"Hierarchical topic models and the nested chinese restaurant process","volume-title":"Advances in Neural Information Processing Systems","author":"Griffiths","year":"2004"},{"key":"2021091013024592300_bib25","article-title":"The curious case of neural text degeneration","volume-title":"International Conference on Learning Representations","author":"Holtzman","year":"2019"},{"key":"2021091013024592300_bib26","first-page":"1587","article-title":"Toward controlled generation of text","volume-title":"Proceedings of the 34th International Conference on Machine Learning","author":"Zhiting","year":"2017"},{"key":"2021091013024592300_bib27","doi-asserted-by":"publisher","first-page":"2101","DOI":"10.18653\/v1\/D17-1223","article-title":"Extractive summarization using multi-task learning with document classification","volume-title":"Proceedings of the 2017 Conference on empirical methods in natural language processing","author":"Isonuma","year":"2017"},{"key":"2021091013024592300_bib28","doi-asserted-by":"publisher","first-page":"800","DOI":"10.18653\/v1\/2020.acl-main.73","article-title":"Tree- structured neural topic model","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Isonuma","year":"2020"},{"key":"2021091013024592300_bib29","doi-asserted-by":"publisher","first-page":"2142","DOI":"10.18653\/v1\/P19-1206","article-title":"Unsupervised neural single-document summarization of reviews via learning latent discourse structure and its ranking","volume-title":"Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics","author":"Isonuma","year":"2019"},{"key":"2021091013024592300_bib30","article-title":"Categorical reparameterization with gumbel-softmax","volume-title":"Proceedings of the 5th International Conference on Learning Representations","author":"Jang","year":"2017"},{"key":"2021091013024592300_bib31","doi-asserted-by":"crossref","first-page":"783","DOI":"10.1145\/2396761.2396861","article-title":"Modeling topic hierarchies with the recursive chinese restaurant process","volume-title":"Proceedings of the 21st ACM International Conference on Information and Knowledge Management","author":"Kim","year":"2012"},{"key":"2021091013024592300_bib32","article-title":"Adam: A method for stochastic optimization","volume":"arXiv:1412.6980v9","author":"Kingma","year":"2014","journal-title":"CoRR"},{"key":"2021091013024592300_bib33","article-title":"Auto-encoding variational Bayes","volume-title":"Proceedings of the 2nd International Conference on Learning Representations","author":"Kingma","year":"2014"},{"key":"2021091013024592300_bib34","doi-asserted-by":"publisher","first-page":"811","DOI":"10.18653\/v1\/N16-1095","article-title":"Capturing reliable fine-grained sentiment associations by crowdsourcing and best\u2013worst scaling","volume-title":"Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","author":"Kiritchenko","year":"2016"},{"key":"2021091013024592300_bib35","doi-asserted-by":"publisher","first-page":"6610","DOI":"10.1609\/aaai.v33i01.33016610","article-title":"Domain agnostic real-valued specificity prediction","volume-title":"Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence","author":"Ko","year":"2019"},{"key":"2021091013024592300_bib36","article-title":"ALBERT: A lite BERT for self-supervised learning of language representations","volume-title":"Proceedings of the 7th International Conference on Learning Representations","author":"Lan","year":"2019"},{"key":"2021091013024592300_bib37","article-title":"ROUGE: A package for automatic evaluation of summaries","volume-title":"Proceedings of the Workshop on Text Summarization Branches Out","author":"Lin","year":"2004"},{"key":"2021091013024592300_bib38","doi-asserted-by":"publisher","first-page":"5070","DOI":"10.18653\/v1\/P19-1500","article-title":"Hierarchical transformers for multi-document summarization","volume-title":"Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics","author":"Liu","year":"2019"},{"key":"2021091013024592300_bib39","doi-asserted-by":"publisher","first-page":"3721","DOI":"10.18653\/v1\/D19-1387","article-title":"Text summarization with pretrained encoders","volume-title":"Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing","author":"Liu","year":"2019"},{"key":"2021091013024592300_bib40","first-page":"605","article-title":"Automatic identification of general and specific sentences by leveraging discourse annotations","volume-title":"Proceedings of 5th International Joint Conference on Natural Language Processing","author":"Louis","year":"2011"},{"key":"2021091013024592300_bib41","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781107337855","volume-title":"Best-Worst Scaling: Theory, Methods and Applications","author":"Louviere","year":"2015"},{"key":"2021091013024592300_bib42","doi-asserted-by":"publisher","first-page":"1412","DOI":"10.18653\/v1\/D15-1166","article-title":"Effective approaches to attention-based neural machine translation","volume-title":"Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing","author":"Luong","year":"2015"},{"key":"2021091013024592300_bib43","article-title":"The concrete distribution: A continuous relaxation of discrete random variables","volume-title":"Proceedings of the 5th International Conference on Learning Representations","author":"Maddison","year":"2017"},{"key":"2021091013024592300_bib44","doi-asserted-by":"publisher","first-page":"1906","DOI":"10.18653\/v1\/2020.acl-main.173","article-title":"On faithfulness and factuality in abstractive summarization","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Maynez","year":"2020"},{"key":"2021091013024592300_bib45","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1145\/2766462.2767755","article-title":"Image-based recommendations on styles and substitutes","volume-title":"Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval","author":"McAuley","year":"2015"},{"key":"2021091013024592300_bib46","first-page":"2410","article-title":"Discovering discrete latent topics with neural variational inference","volume-title":"Proceedings of the 34th International Conference on Machine Learning","author":"Miao","year":"2017"},{"key":"2021091013024592300_bib47","doi-asserted-by":"publisher","first-page":"280","DOI":"10.18653\/v1\/K16-1028","article-title":"Abstractive text summarization using sequence-to-sequence RNNs and beyond","volume-title":"Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning","author":"Nallapati","year":"2016"},{"issue":"2","key":"2021091013024592300_bib48","doi-asserted-by":"publisher","first-page":"256","DOI":"10.1109\/TPAMI.2014.2318728","article-title":"Nested hierarchical Dirichlet processes","volume":"37","author":"Paisley","year":"2014","journal-title":"IEEE Transactions on Pattern Analysis and Machine Intelligence"},{"key":"2021091013024592300_bib49","first-page":"1278","article-title":"Stochastic backpropagation and approximate inference in deep generative models","volume-title":"Proceedings of the 31st International Conference on Machine Learning","author":"Rezende","year":"2014"},{"key":"2021091013024592300_bib50","first-page":"1073","article-title":"Get to the point: Summarization with pointer-generator networks","volume-title":"Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics","author":"See","year":"2017"},{"key":"2021091013024592300_bib51","doi-asserted-by":"publisher","first-page":"5093","DOI":"10.18653\/v1\/D19-1513","article-title":"A topic augmented text generation model: Joint learning of semantics and structural features","volume-title":"Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing","author":"Tang","year":"2019"},{"key":"2021091013024592300_bib52","first-page":"308","article-title":"A joint model of text and aspect ratings for sentiment summarization","volume-title":"Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies","author":"Titov","year":"2008"},{"key":"2021091013024592300_bib53","article-title":"Word representations via Gaussian embedding","volume-title":"Proceedings of the 3rd International Conference on Learning Representations","author":"Vilnis","year":"2015"},{"key":"2021091013024592300_bib54","doi-asserted-by":"publisher","first-page":"166","DOI":"10.18653\/v1\/N19-1015","article-title":"Topic-guided variational auto-encoder for text generation","volume-title":"Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","author":"Wang","year":"2019"},{"key":"2021091013024592300_bib55","first-page":"3881","article-title":"Improved variational autoencoders for text modeling using dilated convolutions","volume-title":"Proceedings of the 34th International Conference on Machine Learning","author":"Yang","year":"2017"},{"key":"2021091013024592300_bib56","first-page":"1097","article-title":"Texygen: A benchmarking platform for text generation models","volume-title":"Proceedings of the 41st International ACM SIGIR Conference on Research and Development in Information Retrieval","author":"Zhu","year":"2018"}],"container-title":["Transactions of the Association for Computational Linguistics"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00406\/1962464\/tacl_a_00406.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"http:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00406\/1962464\/tacl_a_00406.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2021,9,20]],"date-time":"2021-09-20T15:24:42Z","timestamp":1632151482000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/tacl\/article\/doi\/10.1162\/tacl_a_00406\/107204\/Unsupervised-Abstractive-Opinion-Summarization-by"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021]]},"references-count":56,"URL":"https:\/\/doi.org\/10.1162\/tacl_a_00406","relation":{},"ISSN":["2307-387X"],"issn-type":[{"value":"2307-387X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2021]]},"published":{"date-parts":[[2021]]}}}