[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-1-en-105":3,"doc-seo-571631-105":53,"doc-detail-571631-en":126},{"code":4,"msg":5,"data":6},0,"success",[7,14,19,24,29,34,39,44,49],{"id":8,"doc_module":9,"doc_module_name":10,"category_name":11,"show_sort_weight":12,"slug":13},11,1,"Template","Presentations",90,"presentations",{"id":15,"doc_module":9,"doc_module_name":10,"category_name":16,"show_sort_weight":17,"slug":18},12,"Resumes",80,"resumes",{"id":20,"doc_module":9,"doc_module_name":10,"category_name":21,"show_sort_weight":22,"slug":23},14,"Invoices",70,"invoices",{"id":25,"doc_module":9,"doc_module_name":10,"category_name":26,"show_sort_weight":27,"slug":28},15,"Posters",60,"posters",{"id":30,"doc_module":9,"doc_module_name":10,"category_name":31,"show_sort_weight":32,"slug":33},16,"Social Media",50,"social-media",{"id":35,"doc_module":9,"doc_module_name":10,"category_name":36,"show_sort_weight":37,"slug":38},17,"Forms",40,"forms",{"id":40,"doc_module":9,"doc_module_name":10,"category_name":41,"show_sort_weight":42,"slug":43},18,"Letters",30,"letters",{"id":45,"doc_module":9,"doc_module_name":10,"category_name":46,"show_sort_weight":47,"slug":48},21,"Paper Templates",5,"papers-templates",{"id":50,"doc_module":9,"doc_module_name":10,"category_name":51,"show_sort_weight":4,"slug":52},158,"General","general-158",{"code":4,"msg":54,"data":55},"ok",{"site_id":56,"language":57,"slug":58,"title":59,"keywords":60,"description":61,"schema_data":62,"social_meta":119,"head_meta":121,"extra_data":123,"updated_unix":125},105,"en","discosum-discourse-aware-news-summarization-algorithm-and-dataset-for-structure-aware-news-summaries","DiscoSum: Discourse-aware News Summarization - Algorithm and dataset for structure-aware news summaries","","DiscoSum presents discourse-structure-aware news summarization that addresses how large language models often fail to maintain long-range discourse structure, especially in news where narrative flow shapes reader engagement. The work introduces a cross-platform dataset of 20k news articles with multiple human-written summaries over 4 platforms and proposes a discourse schema with five sentence-level discourse labels. A beam-search decoding method selects structure-consistent subsequent sentences, with human and automatic evaluations showing improved narrative fidelity and structural compliance.",{"@graph":63,"@context":118},[64,80,101],{"@type":65,"itemListElement":66},"BreadcrumbList",[67,71,74,77],{"item":68,"name":69,"@type":70,"position":9},"https://docshare.wps.com","Home","ListItem",{"item":72,"name":10,"@type":70,"position":73},"https://docshare.wps.com/template/",2,{"item":75,"name":51,"@type":70,"position":76},"https://docshare.wps.com/template/general/",3,{"item":78,"name":59,"@type":70,"position":79},"https://docshare.wps.com/template/discosum-discourse-aware-news-summarization-algorithm-and-dataset-for-structure-aware-news-summaries/571631/",4,{"url":78,"name":59,"@type":81,"image":82,"author":87,"headline":59,"publisher":90,"fileFormat":93,"inLanguage":57,"description":61,"dateModified":94,"datePublished":95,"encodingFormat":93,"isAccessibleForFree":96,"interactionStatistic":97},"DigitalDocument",{"url":83,"@type":84,"width":85,"height":86},"https://docshare.wps.com/thumbnails/discosum-discourse-aware-news-summarization-algorithm-and-dataset-for-structure-aware-news-summaries/571631.png","ImageObject",442,249,{"name":88,"@type":89},"CatatanPagi","Person",{"url":68,"name":91,"@type":92},"DocShare","Organization","application/pdf","2026-10-21","2026-10-13",true,{"@type":98,"interactionType":99,"userInteractionCount":73},"InteractionCounter",{"@type":100},"ViewAction",{"@type":102,"mainEntity":103},"FAQPage",[104,110,114],{"name":105,"@type":106,"acceptedAnswer":107},"What problem does DiscoSum target in news summarization?","Question",{"text":108,"@type":109},"It targets the tendency of language models to lose long-term discourse structure in summaries, which affects how news narrative flow is conveyed to readers.","Answer",{"name":111,"@type":106,"acceptedAnswer":112},"What dataset does DiscoSum introduce?",{"text":113,"@type":109},"DiscoSum introduces a cross-platform dataset with 20k news articles and over 100k human-written summaries across platforms such as Facebook, Instagram, Twitter, and newsletters.",{"name":115,"@type":106,"acceptedAnswer":116},"How does DiscoSum generate structure-aware summaries?",{"text":117,"@type":109},"It uses a discourse schema with five sentence-level labels and a beam-search decoding method to choose the most suitable next sentences while meeting structural requirements.","https://schema.org",{"og:url":78,"og:type":120,"og:title":59,"og:site_name":91,"og:description":61},"article",{"robots":122,"canonical":78},"index,follow",{"doc_id":124,"site_id":56},571631,1792589270,{"code":4,"msg":5,"data":127},{"doc_id":124,"user_id":128,"nickname":88,"user_avatar":129,"doc_module":9,"category_id":50,"category_name":51,"doc_title":59,"doc_description":61,"doc_content":130,"file_id":131,"file_url":132,"file_type":133,"file_size":134,"view_count":73,"is_deleted":4,"is_public":9,"is_downloadable":9,"audit_status":9,"page_count":135,"language":136,"language_code":57,"site_id":56,"html_lang":57,"table_of_contents":137,"faqs":138,"seo_title":139,"seo_description":61,"update_tm":140,"read_time":141},962090894170,"https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","DiscoSum: Discourse-aware News Summarization  \nAlexander Spangher* 1 , Tenghao Huang∗1, Jialiang Gu∗2, Jiatong Shi3 , and Muhao Chen4  \n1University of Southern California Information Sciences Institute  \n2 School of Computer Science, Wuhan University  \n3Independent Contributor  \n4University of California, Davis  \n[spangher@usc.edu](spangher@usc.edu)  \narXiv :2506 .06930v 1 [ cs .CL] 7 Jun 2025  \nAbstract  \nRecent advances in text summarization have predominantly leveraged large language models to generate concise summaries. However, language models often do not maintain longterm discourse structure, especially in news articles, where organizational flow significantly influences reader engagement. We introduce a novel approach to integrating discourse structure into summarization processes, focusing specifically on news articles across various media. We present a novel summarization dataset where news articles are summarized multiple times in different ways across different social media platforms (e.g. LinkedIn, Facebook, etc.) . We develop a novel news discourse schema to describe summarization structures and a novel algorithm, DiscoSum, which employs beam search technique for structureaware summarization, enabling the transformation of news stories to meet different stylistic and structural demands. Both human and automatic evaluation results demonstrate the efficacy of our approach in maintaining narrative fidelity and meeting structural requirements.  \n1 Introduction  \nIn recent years, text summarization has seen remarkable advances, fueled by foundational Large Language Models that produce concise, contextrich overviews of lengthy documents (Li and Chaturvedi, 2024 ; Peper et al., 2024 ; Zhang et al., 2024) . Yet despite these gains, current summarization approaches rarely account for a fundamental aspect of textual organization: discourse structure (Cohan et al., 2018) .  \nModern news organizations like the New York Times increasingly publish news summaries in a variety of media (e.g. print newspapers, mobile apps, podcasts, and social media) each with distinct audience expectations and content formats (Kalsnes  \n*Equal contribution.  \nFigure 1: Comparative presentation of the Apollo 11 moon landing news across multiple platforms by The New York Times. This example showcases the diversity in content formatting and language adaptation for different audiences: a detailed traditional print article, a concise and visually-driven Instagram post, and a professionally oriented LinkedIn summary. Each platform reflects specific editorial strategies to engage its unique audience effectively, underlining the importance of discourse-aware news summarization.  \nand Larsson, 2018 ; Ngoc, 2022) . For instance, an outlet like The New York Times may produce a child-friendly podcast edition that uses simplified language and gentler framing, a condensed Instagram version with concise, visually engaging snippets, and a longer, more detailed write-up on LinkedIn or the newspaper’s own website to cater to professional or academic readers. Transforming a single piece of news into multiple styles and lengths, while preserving its core narrative and emphasis, demands nuanced control over discourse structure (Shen et al., 2017 ; Hu et al., 2017) .  \nDespite the growing interest in automated news summarization (See et al., 2017 ; Zhang et al., 2020 ; Beltagy et al., 2020 ; He et al., 2020 ; Zhao et al., 2022b), existing dataset approaches have overlooked this need 1. To bridge these gaps, we pro-  \n1 See Appendix C for a deeper comparison to Grusky et al.(2020) .  \npose a novel discourse-structure-aware summarization task that emphasizes the modeling of structural discourse beyond surface-level summarization coherence or factual correctness.  \nFirst, we introduce DiscoSum: a Discourseaware News Summarization dataset. DiscoSum represents the largest and most diverse collection of professionally-written cross-platform news summaries, comprising 20k n","cbCaiircSJbyOvwT","https://ap.wps.com/l/cbCaiircSJbyOvwT","pdf",1410446,22,"English","# Abstract\n# 1 Introduction\n# 2 Related Works","[{\"question\":\"What problem does DiscoSum target in news summarization?\",\"answer\":\"It targets the tendency of language models to lose long-term discourse structure in summaries, which affects how news narrative flow is conveyed to readers.\"},{\"question\":\"What dataset does DiscoSum introduce?\",\"answer\":\"DiscoSum introduces a cross-platform dataset with 20k news articles and over 100k human-written summaries across platforms such as Facebook, Instagram, Twitter, and newsletters.\"},{\"question\":\"How does DiscoSum generate structure-aware summaries?\",\"answer\":\"It uses a discourse schema with five sentence-level labels and a beam-search decoding method to choose the most suitable next sentences while meeting structural requirements.\"}]","DiscoSum: Discourse-aware News Summarization - Algorithm and dataset for structure-aware news summaries | PDF",1791922816,8]