[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-134839-en":3,"doc-seo-134839-105":31,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},134839,1374391975076,"Riley","https://ap-avatar.wpscdn.com/avatar/14000253ca4ec9f6853?x-image-process=image/resize,m_fixed,w_180,h_180&k=1783305029341752051",8,"Research & Report","Modeling Entity Knowledge for Fact Verification","Fact verification aims to judge the truthfulness of textual claims by retrieving relevant evidence, yet many claims require reasoning over external entity information. This paper proposes a fact verification model that integrates entity knowledge by retrieving descriptive text from Wikipedia for each recognized entity, encoding it with a lightweight module, and feeding it into the main network via unidirectional attention. It further improves performance using auxiliary relatedness between claims and evidence. Experiments on FEVER report a FEVER score of 72.89%.","Modeling Entity Knowledge for Fact Verification  \nYang Liu, Chenguang Zhu, Michael Zeng  \nMicrosoft Cognitive Services Research  \n1 Microsoft Way, Redmond, WA, USA {yaliu10,chezhu,[nzeng}@microsoft.com](nzeng}@microsoft.com)  \nAbstract  \nFact verification is a challenging task of identifying the truthfulness of given claims based on the retrieval of relevant evidence texts. Many claims require understanding and reasoning over external entity information for precise verification. In this paper, we propose a novel fact verification model using entity knowledge to enhance its performance. We retrieve descriptive text from Wikipedia for each entity, and then encode these descriptions by a smaller lightweight network to be fed into the main verification model. Furthermore, we boost model performance by adopting and predicting the relatedness between the claim and each evidence as additional signals. We demonstrate experimentally on a large-scale benchmark dataset FEVER that our framework achieves competitive results with a FEVER score of 72 . 89% on the test set.  \n1 Introduction  \nThe rapid development of online applications provides open and efficient platforms for spreading information. However, false information, including fake news and online rumors, have also been growing and spreading widely over the past several years. Vosoughi et al. (2018) shows that false news travels even faster, deeper and broader than the truth. To prevent harm from this false information, automatically verifying the truthfulness of textual contents is becoming an urgent need for our society. In this work, we study fact verification with the goal of automatically assessing the veracity of a textual claim given supporting evidence.  \nMost existing methods consider fact verificationas a natural language inference task (Angeli and Manning, 2014) . Usually, these systems concatenate claim and its supporting evidence sentences, and then feed them into a classification model (Nie et al., 2019) . Alternatively, previous studies construct graph structures based on claim and evidence,  \nand reason over this graph with graph neural networks (Zhou et al., 2019 ; Liu et al., 2020) or Transformer models (Zhong et al., 2020), which are used in top systems in the FEVER challenge (Thorne et al., 2018) . While these studies focus on reasoning based on claim and evidence text, we believe entity knowledge is also important for precise fact verification. For example, given the first claim from FEVER dataset in Table 1, making the correct verification requires a model to understand what is“Wii U” and “OS X” and know the fact that they are not Microsoft and Sony platforms. Similarly, for the second claim, the knowledge that “New York City” is in United States can also be potentially useful for verifying the claim. This information isnot included in the gold evidence provided by the dataset.  \nIn this work, we present a fact verification model that can effectively incorporate external entity information. Given a claim and its evidence sentences, we first recognize named entities from them, linking them with Wikipedia articles, and then retrieve the lead sections of these articles as the entity descriptions. To make the most of this entity knowledge while not introducing noisy information, we propose a lightweight entity knowledge encoder module for representing external entity knowledge. Our large fact verification network then accesses this knowledge by a unidirectional attention mechanism at each encoding layer. Meanwhile, since the input evidence sentences are obtained by an upstream retrieval module, some evidence maybe irrelevant to the claim. Thus, we predict and adopt this relatedness between each evidence and the claim as an auxiliary signal to train our model.  \nWe experiment with our approach on FEVER (Thorne et al., 2018), one influential benchmark dataset for fact verification. FEVER contains over 185k labeled claims and each verifiable claim is paired with several natural langua","cbCaieOO2Ax5WQCP","https://ap.wps.com/l/cbCaieOO2Ax5WQCP","pdf",418224,3,1,10,"English","en",105,"# Abstract\n# Introduction\n## Problem and motivation\n## Related work\n## Proposed approach\n## Experimental setup and contributions","[{\"question\":\"What is the main goal of the fact verification task in this paper?\",\"answer\":\"It focuses on automatically assessing whether a textual claim is true, using supporting evidence and reasoning over it.\"},{\"question\":\"How does the model use entity knowledge to improve verification?\",\"answer\":\"It recognizes named entities in the claim and evidence, links them to Wikipedia articles, retrieves lead descriptions, encodes them with a lightweight module, and injects this knowledge via unidirectional attention at each encoding layer.\"},{\"question\":\"What auxiliary signal does the paper add during training?\",\"answer\":\"It predicts and adopts the relatedness between each evidence sentence and the claim, helping the model handle potentially irrelevant evidence from the upstream retrieval module.\"}]","Modeling Entity Knowledge for Fact Verification | PDF",1787300242,25,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":29},"modeling-entity-knowledge-for-fact-verification","",{"@graph":37,"@context":86},[38,54,69],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,49,51],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":48},"https://docshare.wps.com/document/","Document",2,{"item":50,"name":12,"@type":44,"position":20},"https://docshare.wps.com/document/research-report/",{"item":52,"name":13,"@type":44,"position":53},"https://docshare.wps.com/document/modeling-entity-knowledge-for-fact-verification/134839/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":42,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-25","2026-08-21",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What is the main goal of the fact verification task in this paper?","Question",{"text":76,"@type":77},"It focuses on automatically assessing whether a textual claim is true, using supporting evidence and reasoning over it.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the model use entity knowledge to improve verification?",{"text":81,"@type":77},"It recognizes named entities in the claim and evidence, links them to Wikipedia articles, retrieves lead descriptions, encodes them with a lightweight module, and injects this knowledge via unidirectional attention at each encoding layer.",{"name":83,"@type":74,"acceptedAnswer":84},"What auxiliary signal does the paper add during training?",{"text":85,"@type":77},"It predicts and adopts the relatedness between each evidence sentence and the claim, helping the model handle potentially irrelevant evidence from the upstream retrieval module.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,135],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":48,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":47,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":47,"category_name":133,"show_sort_weight":22,"slug":134},"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":47,"category_name":137,"show_sort_weight":107,"slug":138},19,"General","general"]