{"id":"https://openalex.org/W7162312847","doi":"https://doi.org/10.48550/arxiv.2605.22829","title":"LFRAG: Layout-oriented Fine-grained Retrieval-Augmented Generation on Multimodal Document Understanding","display_name":"LFRAG: Layout-oriented Fine-grained Retrieval-Augmented Generation on Multimodal Document Understanding","publication_year":2026,"publication_date":"2026-04-18","ids":{"openalex":"https://openalex.org/W7162312847","doi":"https://doi.org/10.48550/arxiv.2605.22829"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.22829","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22829","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.22829","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136982928","display_name":"Yifan Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Yifan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114402760","display_name":"Yu Mi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mi, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136906218","display_name":"Yue Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Yue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102544375","display_name":"Yanchu Guan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guan, Yanchu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136967531","display_name":"Zhixuan Chu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chu, Zhixuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9028000235557556,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9028000235557556,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.050999999046325684,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.007899999618530273,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.6377000212669373},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.5922999978065491},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.5411999821662903},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.5364000201225281},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5307999849319458},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.49309998750686646},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.3603000044822693},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.35519999265670776}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8495000004768372},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.6377000212669373},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.5922999978065491},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.5411999821662903},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5364000201225281},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5307999849319458},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.49309998750686646},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46779999136924744},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.4316999912261963},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.3603000044822693},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.35519999265670776},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.3197999894618988},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.31049999594688416},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.305400013923645},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.29499998688697815},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.2924000024795532},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.28029999136924744},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.2671999931335449},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.25589999556541443}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.22829","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22829","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.22829","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22829","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","score":0.5667248368263245,"display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multimodal":[0],"Retrieval-Augmented":[1,63],"Generation":[2,64],"(RAG)":[3],"has":[4],"emerged":[5],"as":[6,187],"an":[7,188],"effective":[8],"paradigm":[9],"for":[10,118,193],"enhancing":[11],"Large":[12],"Language":[13],"Models":[14],"(LLMs)":[15],"with":[16,99,131,148],"external":[17],"knowledge.":[18],"However,":[19],"existing":[20,152],"multimodal":[21,71,142,194],"RAG":[22,72,195],"systems":[23],"predominantly":[24],"rely":[25],"on":[26,156,164],"coarse-grained":[27],"page-level":[28,74],"retrieval,":[29,108],"which":[30],"fails":[31],"to":[32,49,75,82,139],"capture":[33],"fine-grained":[34,86],"semantic":[35],"and":[36,47,89,114,145,176,190,202],"layout":[37,80],"structures":[38],"in":[39,52,173,182],"visually":[40,197],"rich":[41,198],"documents,":[42],"thereby":[43],"compromising":[44],"retrieval":[45,87,144,165],"accuracy":[46],"leading":[48],"redundant":[50],"context":[51,101],"downstream":[53,119],"tasks.":[54],"To":[55,121],"address":[56],"these":[57],"issues,":[58],"we":[59,125],"propose":[60],"Layout-oriented":[61],"Fine-grained":[62],"(LFRAG),":[65],"a":[66,91,128],"novel":[67],"framework":[68,192],"that":[69,95,159],"advances":[70],"from":[73],"block-level":[76,105,132],"retrieval.":[77],"We":[78],"perform":[79],"segmentation":[81],"construct":[83,126],"semantically":[84],"coherent":[85],"units":[88],"design":[90],"semantic-layout":[92],"fusion":[93],"encoder":[94],"integrates":[96],"local":[97],"semantics":[98],"global":[100],"via":[102],"cross-attention.":[103],"With":[104],"late":[106],"interaction":[107],"LFRAG":[109,160,186],"enables":[110],"precise":[111],"query-content":[112],"alignment":[113],"reduces":[115,177],"irrelevant":[116],"content":[117],"generation.":[120],"enable":[122],"rigorous":[123],"evaluation,":[124],"LFDocQA,":[127],"large-scale":[129],"benchmark":[130],"annotations":[133],"spanning":[134],"diverse":[135],"document":[136,143],"types,":[137],"designed":[138],"assess":[140],"both":[141],"question":[146],"answering":[147],"greater":[149],"granularity":[150],"than":[151],"datasets.":[153],"Extensive":[154],"experiments":[155],"LFDocQA":[157],"demonstrate":[158],"achieves":[161],"state-of-the-art":[162],"performance":[163],"tasks,":[166,184],"outperforms":[167],"the":[168],"best":[169],"baseline":[170],"by":[171,180],"7.20%":[172],"answer":[174],"accuracy,":[175],"token":[178],"consumption":[179],"73.07%":[181],"generation":[183],"confirming":[185],"accurate":[189],"efficient":[191],"over":[196],"documents.":[199],"Our":[200],"code":[201],"datasets":[203],"will":[204],"be":[205],"released":[206],"soon.":[207]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-26T00:00:00"}