[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85938-en":3,"doc-seo-85938-105":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},85938,7971461740909,"Levi","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","EvidentialRAG Quantifying and Mitigating Information Conflict in Multi Source Retrieval Augmented Generation via Evidential Deep Learning","Retrieval-augmented generation grounds large language models in external evidence, yet many pipelines treat retrieved passages as deterministic and mutually consistent. In open information environments, sources can disagree due to temporal drift, source errors, ambiguity, or genuine uncertainty. EvidentialRAG introduces an uncertainty-aware RAG framework that converts retrieved chunks into probabilistic evidence, fuses conflict with Dempster-Shafer methods, and routes generation toward direct answering, conflict-aware answering, or abstention based on fused uncertainty scores. Experiments on CRAG, ConflictQA, and MuSiQue show improved conflict behavior and reduced hallucination under ambiguity, with better calibration.","arXiv :2607 . 1049 1v 1 [ cs .LG] 11 Jul 2026  \nEvidentialRAG: Quantifying and Mitigating Information Conflict in Multi-Source Retrieval-Augmented Generation via Evidential Deep  \nLearning  \nS M Asif Hossaina,, Ruksat Khan Shayonia , M. F. Mridhab  \na School of Computing, Wichita State University, Kansas, USA b Department of Computer Science, American International  \nUniversity-Bangladesh, Dhaka, Bangladesh  \nAbstract  \nRetrieval-augmented generation grounds large language models in external evidence, but most pipelines still treat retrieved passages as deterministic and mutually consistent context. In open information environments, retrieved sources may disagree because of temporal drift, source error, ambiguity, or genuine uncertainty. This paper introduces EvidentialRAG, an uncertaintyaware RAG framework that converts retrieved chunks into probabilistic evidence before generation. A lightweight evaluator extracts candidate claims and maps chunk-level support to Dirichlet evidence. A conflict-preserving Dempster-Shafer fusion rule then transfers unresolved disagreement into epistemic uncertainty rather than normalizing it away. The generator is routed to direct answering, conflict-aware answering, or abstention according to the fused uncertainty score. Experiments on CRAG, ConflictQA, and MuSiQue show that EvidentialRAG remains competitive with the strongest matched baseline on standard question answering while improving behavior under conflict. On the CRAG ambiguous subset, hallucination decreases from 45.3% for Corrective RAG to a human-calibrated estimate of 34.8%, conflict resolution increases from 35.2% to 51.2%, and expected calibration error improves to 0.122 . These results suggest that evidential modeling is a practical mechanism for trustworthy information processing in foundationmodel-based retrieval systems.  \nKeywords: Retrieval augmented generation, Large language models, Evidential deep learning, Knowledge conflict, Dempster-Shafer theory,  \nHallucination, Calibration  \n1. Introduction  \nLarge language models (LLMs) have transformed knowledge-intensive natural language processing by providing general-purpose capabilities for question answering, summarization, reasoning, and dialogue (Vaswani et al. , 2017 ; Devlin et al. , 2019 ; Brown et al. , 2020 ; OpenAI, 2023 ; Touvron et al. , 2023 ; Grattafiori et al. , 2024) . Their usefulness in information systems is nevertheless constrained by a persistent limitation: model parameters are incomplete, opaque, temporally bounded, and difficult to update. Retrievalaugmented generation (RAG) addresses this limitation by conditioning the generator on externally retrieved documents, passages, database records, or web evidence (Lewis et al. , 2020 ; Guu et al. , 2020 ; Izacard and Grave, 2021 ; Borgeaud et al. , 2022) . By separating knowledge access from language generation, RAG has become a dominant design pattern for enterprise search, digital libraries, domain-specific question answering, and decision-support systems (Gao et al. , 2024 ; Asai et al. , 2024 ; Yan et al. , 2024) .  \nThe conventional RAG pipeline is effective when retrieval returns relevant and mutually consistent evidence. In real information environments, however, retrieval is not only a relevance-ranking problem. It is also an evidence-quality and source-consistency problem. Multi-source retrieval may return outdated statements, near-duplicate claims with small but important differences, contradictory web snippets, partially relevant passages, and information whose applicability depends on time, jurisdiction, source authority, or user intent. This situation is common in healthcare, law, public policy, finance, scientific search, and organizational knowledge management. In such settings, a useful information system should not merely produce a fluent answer. It should identify whether the answer is supported, whether sources disagree, and whether a definitive answer is warranted.  \nCurrent RAG systems of","cbCaimoRXaZIZmxg","https://ap.wps.com/l/cbCaimoRXaZIZmxg","pdf",572603,1,43,"English","en",105,"# Introduction\n## Motivation and limitations of conventional RAG\n## Evidence quality and source consistency challenges\n## Uncertainty and conflict-aware generation approach\n# EvidentialRAG method\n## Probabilistic evidence from retrieved chunks\n## Conflict-preserving fusion and uncertainty routing\n# Experiments and results\n## Benchmarks and comparative performance\n## Hallucination, conflict resolution, and calibration outcomes","[{\"question\":\"What problem does EvidentialRAG target in multi-source RAG pipelines?\",\"answer\":\"It targets information conflict, where retrieved sources disagree due to drift, errors, ambiguity, or true uncertainty. Conventional RAG often concatenates chunks and implicitly assumes consistency, which can lead to unsafe confident answers.\"},{\"question\":\"How does EvidentialRAG convert retrieved passages into a form usable for uncertainty reasoning?\",\"answer\":\"It extracts candidate claims with a lightweight evaluator and maps chunk-level support into Dirichlet evidence, turning retrieved chunks into probabilistic evidence prior to generation.\"},{\"question\":\"How does EvidentialRAG handle disagreement during generation?\",\"answer\":\"It uses a conflict-preserving Dempster-Shafer fusion rule to transfer unresolved disagreement into epistemic uncertainty. The generator then routes to direct answering, conflict-aware answering, or abstention based on the fused uncertainty score.\"}]",1784207261,108,{"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},"evidentialrag-quantifying-and-mitigating-information-conflict-in-multi-source-retrieval-augmented-generation-via-evidential-deep-learning","",{"@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/evidentialrag-quantifying-and-mitigating-information-conflict-in-multi-source-retrieval-augmented-generation-via-evidential-deep-learning/85938/",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},"What problem does EvidentialRAG target in multi-source RAG pipelines?","Question",{"text":75,"@type":76},"It targets information conflict, where retrieved sources disagree due to drift, errors, ambiguity, or true uncertainty. Conventional RAG often concatenates chunks and implicitly assumes consistency, which can lead to unsafe confident answers.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does EvidentialRAG convert retrieved passages into a form usable for uncertainty reasoning?",{"text":80,"@type":76},"It extracts candidate claims with a lightweight evaluator and maps chunk-level support into Dirichlet evidence, turning retrieved chunks into probabilistic evidence prior to generation.",{"name":82,"@type":73,"acceptedAnswer":83},"How does EvidentialRAG handle disagreement during generation?",{"text":84,"@type":76},"It uses a conflict-preserving Dempster-Shafer fusion rule to transfer unresolved disagreement into epistemic uncertainty. The generator then routes to direct answering, conflict-aware answering, or abstention based on the fused uncertainty score.","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},[93,97,101,105,110,115,120,123,128,131,135],{"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":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]