Loading…

RankSum—An unsupervised extractive text summarization based on rank fusion

In this paper, we propose Ranksum, an approach for extractive text summarization of single documents based on the rank fusion of four multi-dimensional sentence features extracted for each sentence: topic information, semantic content, significant keywords, and position. The Ranksum obtains the sent...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications 2022-08, Vol.200, p.116846, Article 116846
Main Authors: Joshi, Akanksha, Fidalgo, Eduardo, Alegre, Enrique, Alaiz-Rodriguez, Rocio
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by cdi_FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103
cites cdi_FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103
container_end_page
container_issue
container_start_page 116846
container_title Expert systems with applications
container_volume 200
creator Joshi, Akanksha
Fidalgo, Eduardo
Alegre, Enrique
Alaiz-Rodriguez, Rocio
description In this paper, we propose Ranksum, an approach for extractive text summarization of single documents based on the rank fusion of four multi-dimensional sentence features extracted for each sentence: topic information, semantic content, significant keywords, and position. The Ranksum obtains the sentence saliency rankings corresponding to each feature in an unsupervised way followed by the weighted fusion of the four scores to rank the sentences according to their significance. The scores are generated in completely unsupervised way, and a labeled document set is required to learn the fusion weights. Since we found that the fusion weights can generalize to other datasets, we consider the Ranksum as an unsupervised approach. To determine topic rank, we employ probabilistic topic models whereas semantic information is captured using sentence embeddings. To derive rankings using sentence embeddings, we utilize Siamese networks to produce abstractive sentence representation and then we formulate a novel strategy to arrange them in their order of importance. A graph-based strategy is applied to find the significant keywords and related sentence rankings in the document. We also formulate a sentence novelty measure based on bigrams, trigrams, and sentence embeddings to eliminate redundant sentences from the summary. The ranks of all the sentences – computed for each feature – are finally fused to get the final score for each sentence in the document. We evaluate our approach on publicly available summarization datasets — CNN/DailyMail and DUC 2002. Experimental results show that our approach outperforms other existing state-of-the-art summarization methods. [Display omitted] •A unified summarization framework with multi-dimensional sentence features.•Novel method to rank sentences using probabilistic topic vectors.•Sentences ranked with a new technique exploiting embeddings.•Novelty parameter utilizing bigrams, trigrams and sentence embeddings.•Rank-based fusion strategy for extractive summarization.
doi_str_mv 10.1016/j.eswa.2022.116846
format article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2673376703</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0957417422002998</els_id><sourcerecordid>2673376703</sourcerecordid><originalsourceid>FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103</originalsourceid><addsrcrecordid>eNp9kMtKxDAUhoMoOI6-gKuC69Zc2iQFN8PgDQYEL-uQpqeQ6rRj0tTLyofwCX0SU-ra1fk5_N-5_AidEpwRTPh5m4F_0xnFlGaEcJnzPbQgUrCUi5LtowUuC5HmROSH6Mj7FmMiMBYLtLnX3fND2P58fa-6JHQ-7MCN1kOdwPvgtBnsCMkQdeLDdqud_dSD7buk0pMnChcHJE3wsXmMDhr94uHkry7R09Xl4_om3dxd365Xm9QwKoeUA2eEF0VR1YWWphSY1FVeypLWTMvccFmJSgOW1DSQm7KUrM41kVKKiBLMluhsnrtz_WsAP6i2D66LKxXlgjHBBWbRRWeXcb33Dhq1czZ-8KEIVlNqqlVTampKTc2pRehihiDeP1pwyhsLnYHaOjCDqnv7H_4Lp7t24g</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2673376703</pqid></control><display><type>article</type><title>RankSum—An unsupervised extractive text summarization based on rank fusion</title><source>Elsevier:Jisc Collections:Elsevier Read and Publish Agreement 2022-2024:Freedom Collection (Reading list)</source><creator>Joshi, Akanksha ; Fidalgo, Eduardo ; Alegre, Enrique ; Alaiz-Rodriguez, Rocio</creator><creatorcontrib>Joshi, Akanksha ; Fidalgo, Eduardo ; Alegre, Enrique ; Alaiz-Rodriguez, Rocio</creatorcontrib><description>In this paper, we propose Ranksum, an approach for extractive text summarization of single documents based on the rank fusion of four multi-dimensional sentence features extracted for each sentence: topic information, semantic content, significant keywords, and position. The Ranksum obtains the sentence saliency rankings corresponding to each feature in an unsupervised way followed by the weighted fusion of the four scores to rank the sentences according to their significance. The scores are generated in completely unsupervised way, and a labeled document set is required to learn the fusion weights. Since we found that the fusion weights can generalize to other datasets, we consider the Ranksum as an unsupervised approach. To determine topic rank, we employ probabilistic topic models whereas semantic information is captured using sentence embeddings. To derive rankings using sentence embeddings, we utilize Siamese networks to produce abstractive sentence representation and then we formulate a novel strategy to arrange them in their order of importance. A graph-based strategy is applied to find the significant keywords and related sentence rankings in the document. We also formulate a sentence novelty measure based on bigrams, trigrams, and sentence embeddings to eliminate redundant sentences from the summary. The ranks of all the sentences – computed for each feature – are finally fused to get the final score for each sentence in the document. We evaluate our approach on publicly available summarization datasets — CNN/DailyMail and DUC 2002. Experimental results show that our approach outperforms other existing state-of-the-art summarization methods. [Display omitted] •A unified summarization framework with multi-dimensional sentence features.•Novel method to rank sentences using probabilistic topic vectors.•Sentences ranked with a new technique exploiting embeddings.•Novelty parameter utilizing bigrams, trigrams and sentence embeddings.•Rank-based fusion strategy for extractive summarization.</description><identifier>ISSN: 0957-4174</identifier><identifier>EISSN: 1873-6793</identifier><identifier>DOI: 10.1016/j.eswa.2022.116846</identifier><language>eng</language><publisher>New York: Elsevier Ltd</publisher><subject>Artificial neural networks ; Datasets ; Documents ; Embeddings ; Extractive ; Feature extraction ; Keywords ; Semantics ; Sentences ; Text summarization ; Topic</subject><ispartof>Expert systems with applications, 2022-08, Vol.200, p.116846, Article 116846</ispartof><rights>2022 Elsevier Ltd</rights><rights>Copyright Elsevier BV Aug 15, 2022</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103</citedby><cites>FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103</cites><orcidid>0000-0003-1202-5232 ; 0000-0002-6107-8595</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids></links><search><creatorcontrib>Joshi, Akanksha</creatorcontrib><creatorcontrib>Fidalgo, Eduardo</creatorcontrib><creatorcontrib>Alegre, Enrique</creatorcontrib><creatorcontrib>Alaiz-Rodriguez, Rocio</creatorcontrib><title>RankSum—An unsupervised extractive text summarization based on rank fusion</title><title>Expert systems with applications</title><description>In this paper, we propose Ranksum, an approach for extractive text summarization of single documents based on the rank fusion of four multi-dimensional sentence features extracted for each sentence: topic information, semantic content, significant keywords, and position. The Ranksum obtains the sentence saliency rankings corresponding to each feature in an unsupervised way followed by the weighted fusion of the four scores to rank the sentences according to their significance. The scores are generated in completely unsupervised way, and a labeled document set is required to learn the fusion weights. Since we found that the fusion weights can generalize to other datasets, we consider the Ranksum as an unsupervised approach. To determine topic rank, we employ probabilistic topic models whereas semantic information is captured using sentence embeddings. To derive rankings using sentence embeddings, we utilize Siamese networks to produce abstractive sentence representation and then we formulate a novel strategy to arrange them in their order of importance. A graph-based strategy is applied to find the significant keywords and related sentence rankings in the document. We also formulate a sentence novelty measure based on bigrams, trigrams, and sentence embeddings to eliminate redundant sentences from the summary. The ranks of all the sentences – computed for each feature – are finally fused to get the final score for each sentence in the document. We evaluate our approach on publicly available summarization datasets — CNN/DailyMail and DUC 2002. Experimental results show that our approach outperforms other existing state-of-the-art summarization methods. [Display omitted] •A unified summarization framework with multi-dimensional sentence features.•Novel method to rank sentences using probabilistic topic vectors.•Sentences ranked with a new technique exploiting embeddings.•Novelty parameter utilizing bigrams, trigrams and sentence embeddings.•Rank-based fusion strategy for extractive summarization.</description><subject>Artificial neural networks</subject><subject>Datasets</subject><subject>Documents</subject><subject>Embeddings</subject><subject>Extractive</subject><subject>Feature extraction</subject><subject>Keywords</subject><subject>Semantics</subject><subject>Sentences</subject><subject>Text summarization</subject><subject>Topic</subject><issn>0957-4174</issn><issn>1873-6793</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><recordid>eNp9kMtKxDAUhoMoOI6-gKuC69Zc2iQFN8PgDQYEL-uQpqeQ6rRj0tTLyofwCX0SU-ra1fk5_N-5_AidEpwRTPh5m4F_0xnFlGaEcJnzPbQgUrCUi5LtowUuC5HmROSH6Mj7FmMiMBYLtLnX3fND2P58fa-6JHQ-7MCN1kOdwPvgtBnsCMkQdeLDdqud_dSD7buk0pMnChcHJE3wsXmMDhr94uHkry7R09Xl4_om3dxd365Xm9QwKoeUA2eEF0VR1YWWphSY1FVeypLWTMvccFmJSgOW1DSQm7KUrM41kVKKiBLMluhsnrtz_WsAP6i2D66LKxXlgjHBBWbRRWeXcb33Dhq1czZ-8KEIVlNqqlVTampKTc2pRehihiDeP1pwyhsLnYHaOjCDqnv7H_4Lp7t24g</recordid><startdate>20220815</startdate><enddate>20220815</enddate><creator>Joshi, Akanksha</creator><creator>Fidalgo, Eduardo</creator><creator>Alegre, Enrique</creator><creator>Alaiz-Rodriguez, Rocio</creator><general>Elsevier Ltd</general><general>Elsevier BV</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><orcidid>https://orcid.org/0000-0003-1202-5232</orcidid><orcidid>https://orcid.org/0000-0002-6107-8595</orcidid></search><sort><creationdate>20220815</creationdate><title>RankSum—An unsupervised extractive text summarization based on rank fusion</title><author>Joshi, Akanksha ; Fidalgo, Eduardo ; Alegre, Enrique ; Alaiz-Rodriguez, Rocio</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Artificial neural networks</topic><topic>Datasets</topic><topic>Documents</topic><topic>Embeddings</topic><topic>Extractive</topic><topic>Feature extraction</topic><topic>Keywords</topic><topic>Semantics</topic><topic>Sentences</topic><topic>Text summarization</topic><topic>Topic</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Joshi, Akanksha</creatorcontrib><creatorcontrib>Fidalgo, Eduardo</creatorcontrib><creatorcontrib>Alegre, Enrique</creatorcontrib><creatorcontrib>Alaiz-Rodriguez, Rocio</creatorcontrib><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts – Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>Expert systems with applications</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Joshi, Akanksha</au><au>Fidalgo, Eduardo</au><au>Alegre, Enrique</au><au>Alaiz-Rodriguez, Rocio</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>RankSum—An unsupervised extractive text summarization based on rank fusion</atitle><jtitle>Expert systems with applications</jtitle><date>2022-08-15</date><risdate>2022</risdate><volume>200</volume><spage>116846</spage><pages>116846-</pages><artnum>116846</artnum><issn>0957-4174</issn><eissn>1873-6793</eissn><abstract>In this paper, we propose Ranksum, an approach for extractive text summarization of single documents based on the rank fusion of four multi-dimensional sentence features extracted for each sentence: topic information, semantic content, significant keywords, and position. The Ranksum obtains the sentence saliency rankings corresponding to each feature in an unsupervised way followed by the weighted fusion of the four scores to rank the sentences according to their significance. The scores are generated in completely unsupervised way, and a labeled document set is required to learn the fusion weights. Since we found that the fusion weights can generalize to other datasets, we consider the Ranksum as an unsupervised approach. To determine topic rank, we employ probabilistic topic models whereas semantic information is captured using sentence embeddings. To derive rankings using sentence embeddings, we utilize Siamese networks to produce abstractive sentence representation and then we formulate a novel strategy to arrange them in their order of importance. A graph-based strategy is applied to find the significant keywords and related sentence rankings in the document. We also formulate a sentence novelty measure based on bigrams, trigrams, and sentence embeddings to eliminate redundant sentences from the summary. The ranks of all the sentences – computed for each feature – are finally fused to get the final score for each sentence in the document. We evaluate our approach on publicly available summarization datasets — CNN/DailyMail and DUC 2002. Experimental results show that our approach outperforms other existing state-of-the-art summarization methods. [Display omitted] •A unified summarization framework with multi-dimensional sentence features.•Novel method to rank sentences using probabilistic topic vectors.•Sentences ranked with a new technique exploiting embeddings.•Novelty parameter utilizing bigrams, trigrams and sentence embeddings.•Rank-based fusion strategy for extractive summarization.</abstract><cop>New York</cop><pub>Elsevier Ltd</pub><doi>10.1016/j.eswa.2022.116846</doi><orcidid>https://orcid.org/0000-0003-1202-5232</orcidid><orcidid>https://orcid.org/0000-0002-6107-8595</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 0957-4174
ispartof Expert systems with applications, 2022-08, Vol.200, p.116846, Article 116846
issn 0957-4174
1873-6793
language eng
recordid cdi_proquest_journals_2673376703
source Elsevier:Jisc Collections:Elsevier Read and Publish Agreement 2022-2024:Freedom Collection (Reading list)
subjects Artificial neural networks
Datasets
Documents
Embeddings
Extractive
Feature extraction
Keywords
Semantics
Sentences
Text summarization
Topic
title RankSum—An unsupervised extractive text summarization based on rank fusion
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T15%3A30%3A45IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=RankSum%E2%80%94An%20unsupervised%20extractive%20text%20summarization%20based%20on%20rank%20fusion&rft.jtitle=Expert%20systems%20with%20applications&rft.au=Joshi,%20Akanksha&rft.date=2022-08-15&rft.volume=200&rft.spage=116846&rft.pages=116846-&rft.artnum=116846&rft.issn=0957-4174&rft.eissn=1873-6793&rft_id=info:doi/10.1016/j.eswa.2022.116846&rft_dat=%3Cproquest_cross%3E2673376703%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c328t-6e6316555bd5a8c9701db49892d3a84c68b7bae082cfe4c9983d4a188876e6103%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2673376703&rft_id=info:pmid/&rfr_iscdi=true