[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85499-en":3,"doc-seo-85499-105":29,"detail-sidebar-cat-0-en-105":91},{"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":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},85499,34359740700684,"Finn","https://ap-avatar.wpscdn.com/avatar/1f400023980c374ae676?_k=1777273430885731487",8,"Research & Report","Multimodal Rumor Detection Enhanced by External Evidence and Forgery Features","Social media spreads information through mixed image–text posts, yet rumors exploit subtle inconsistencies and forged content, making detection based only on post content difficult. Deep semantic-mismatch rumors that appear aligned across modalities remain especially challenging and threaten public opinion. The work proposes a multimodal rumor detection model enhanced with external evidence and forgery features, extracting frequency-domain traces and compression artifacts while using ResNet34 and BERT encoders.","Multimodal rumor detection enhanced by external evidence and forgery features  \nHan Li1, Hua Sun2,*  \n1 Information Engineering School of Dalian Ocean University,Dalian Liaoning, China  \n2 Information Engineering School of Dalian Ocean University,Dalian Liaoning, China  \n2,*Corresponding Author: Hua Sun. Email: [sunhua@dlou.edu.cn](sunhua@dlou.edu.cn)  \nABSTRACT: Social media increasingly disseminates information through mixed image–text posts, but rumors often exploit subtle inconsistencies and forged content, making detection based solely on post content difficult. Deep semantic-mismatch rumors, which superficially align images and texts, pose particular challenges and threaten online public opinion. Existing multimodal rumor detection methods improve cross-modal modeling but suffer from limited feature extraction, noisy alignment, and inflexible fusion strategies, while ignoring external factual evidence necessary for verifying complex rumors. To address these limitations, we propose a multimodal rumor detection model enhanced with external evidence and forgery features. The model uses a ResNet34 visual encoder, a BERT text encoder, and a forgery-feature module extracting frequency-domain traces and compression artifacts via Fourier transformation. While some existing approaches employ large-scale generative vision-language models for caption generation, their open-ended generation tendency produces verbose and stylistically inconsistent descriptions that introduce semantic noise and risk drifting from actual image content. To overcome this, we adopt BLIP—specifically pre-trained for vision-language alignment—which generates concise, image-faithful descriptions stylistically closer to news text, serving as a reliable semantic bridge across modalities. A BLIP-Driven Semantic Alignment Module jointly optimizes text– image and text–description contrastive losses, capturing inconsistencies at both visual and semantic levels. A gated adaptive feature-scaling fusion mechanism dynamically adjusts multimodal fusion and reduces redundancy. Experiments on Weibo and Twitter datasets demonstrate that our model outperforms mainstream baselines in M_Acc, recall, and F1 score.  \nKEYWORDS: Multimodal rumor detection; modality fusion; attention mechanism; image forgery  \n1 Introduction  \nThe rapid development of social networks has enabled various types of information to be widely disseminated within an extremely short period of time. While this brings great convenience to users in accessing and sharing information, it also provides favorable conditions for the generation and spread of false content. Driven by the traffic economy, online rumors have significantly improved in terms of content breadth, dissemination speed, update frequency, and scope of influence, which not only disrupt the order of the online space but also erode social trust, posing a serious threat to public interests and social stability.  \nFor example, in the special rectification campaigns recently carried out by relevant authorities, a large number of illegal accounts involved in rumors and massive amounts of rumor-related information have been handled in accordance with the law. These measures fully highlight the urgency and necessity of strengthening rumor governance in the social network environment. Meanwhile, with the continuous diversification of communication forms, postson social platforms are no longer limited to text and often include images to enhance persuasiveness. Rumor makers frequently combine real or synthetic images with false text to form more deceptive image-text combinations (Tufchi et al.,2023) , which makes it more difficult for users to distinguish the authenticity of information and further increases the difficulty of rumor detection.  \nTherefore, authenticity detection for multimodal posts on social networks, aiming to achieve accurate rumor identification and curb the spread of false information, has become a current research hotspot in the f","cbCain71OQPh2eef","https://ap.wps.com/l/cbCain71OQPh2eef","pdf",1067495,1,20,"English","en",105,"# Introduction\n## Multimodal posts and rumor risks\n## Limitations of existing multimodal rumor detection\n## Frequency-domain forgery traces and alignment gaps\n## Social context modeling challenges","[{\"question\":\"Why are multimodal rumors difficult to detect in image–text social posts?\",\"answer\":\"Rumors can combine subtle visual tampering with misleading text, creating deep semantic mismatches that still look cross-modal consistent, which reduces the effectiveness of content-only detection.\"},{\"question\":\"What forgery-related features does the proposed model use?\",\"answer\":\"It includes a forgery-feature module that extracts frequency-domain traces and compression artifacts using Fourier transformation to capture tampering clues often missed by spatial-domain methods.\"},{\"question\":\"How does the model improve semantic alignment across image and text?\",\"answer\":\"It adopts a BLIP-driven semantic alignment module that generates concise, image-faithful descriptions, then jointly optimizes text–image and text–description contrastive losses to capture both visual and semantic inconsistencies.\"}]",1784204031,50,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"multimodal-rumor-detection-enhanced-by-external-evidence-and-forgery-features","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/multimodal-rumor-detection-enhanced-by-external-evidence-and-forgery-features/85499/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why are multimodal rumors difficult to detect in image–text social posts?","Question",{"text":75,"@type":76},"Rumors can combine subtle visual tampering with misleading text, creating deep semantic mismatches that still look cross-modal consistent, which reduces the effectiveness of content-only detection.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What forgery-related features does the proposed model use?",{"text":80,"@type":76},"It includes a forgery-feature module that extracts frequency-domain traces and compression artifacts using Fourier transformation to capture tampering clues often missed by spatial-domain methods.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the model improve semantic alignment across image and text?",{"text":84,"@type":76},"It adopts a BLIP-driven semantic alignment module that generates concise, image-faithful descriptions, then jointly optimizes text–image and text–description contrastive losses to capture both visual and semantic inconsistencies.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":28,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":21,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":21,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":21,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]