{"id":"https://openalex.org/W7171539638","doi":"https://doi.org/10.48550/arxiv.2607.24554","title":"DeCoRAG: Cognitive Decoupling and Semantic-Aware Cropping for Complex Document Understanding","display_name":"DeCoRAG: Cognitive Decoupling and Semantic-Aware Cropping for Complex Document Understanding","publication_year":2026,"publication_date":"2026-07-27","ids":{"openalex":"https://openalex.org/W7171539638","doi":"https://doi.org/10.48550/arxiv.2607.24554"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.24554","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.24554","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"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.2607.24554","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5143765844","display_name":"Shuo Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Shuo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143769913","display_name":"Kai Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Kai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143864455","display_name":"Wenyuan Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Wenyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143780814","display_name":"Yizheng Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Yizheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143840945","display_name":"Xia Liao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liao, Xia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008564177","display_name":"Junming Su","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Su, Junming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143819839","display_name":"Qing Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Qing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5010754242","display_name":"Xi Fang","orcid":"https://orcid.org/0000-0002-3783-2041"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xi, Fang","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":0.650600016117096,"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":0.650600016117096,"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":0.11309999972581863,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.043800000101327896,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/semantics","display_name":"Semantics (computer science)","score":0.5454999804496765},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.48159998655319214},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.44699999690055847},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.40470001101493835},{"id":"https://openalex.org/keywords/semantic-memory","display_name":"Semantic memory","score":0.396699994802475},{"id":"https://openalex.org/keywords/defeasible-estate","display_name":"Defeasible estate","score":0.3517000079154968},{"id":"https://openalex.org/keywords/semantic-compression","display_name":"Semantic compression","score":0.33079999685287476},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.3278999924659729}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7501999735832214},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.5454999804496765},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5231000185012817},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.48159998655319214},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.44699999690055847},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.40470001101493835},{"id":"https://openalex.org/C197914299","wikidata":"https://www.wikidata.org/wiki/Q18650","display_name":"Semantic memory","level":3,"score":0.396699994802475},{"id":"https://openalex.org/C193856179","wikidata":"https://www.wikidata.org/wiki/Q5251100","display_name":"Defeasible estate","level":2,"score":0.3517000079154968},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.34880000352859497},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.33719998598098755},{"id":"https://openalex.org/C202708506","wikidata":"https://www.wikidata.org/wiki/Q7449050","display_name":"Semantic compression","level":5,"score":0.33079999685287476},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.3278999924659729},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.32690000534057617},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.31439998745918274},{"id":"https://openalex.org/C90312973","wikidata":"https://www.wikidata.org/wiki/Q7449052","display_name":"Semantic data model","level":2,"score":0.29280000925064087},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.289000004529953},{"id":"https://openalex.org/C197115733","wikidata":"https://www.wikidata.org/wiki/Q1003136","display_name":"Forcing (mathematics)","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C123606473","wikidata":"https://www.wikidata.org/wiki/Q907918","display_name":"Complex event processing","level":3,"score":0.2870999872684479},{"id":"https://openalex.org/C511149849","wikidata":"https://www.wikidata.org/wiki/Q7449051","display_name":"Semantic computing","level":3,"score":0.26179999113082886},{"id":"https://openalex.org/C193702766","wikidata":"https://www.wikidata.org/wiki/Q1414548","display_name":"Concurrency","level":2,"score":0.26159998774528503},{"id":"https://openalex.org/C99138194","wikidata":"https://www.wikidata.org/wiki/Q183427","display_name":"Hash function","level":2,"score":0.2515000104904175}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.24554","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.24554","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2607.24554","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.24554","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Advancing":[0],"multimodal":[1,52,134],"retrieval-augmented":[2],"generation":[3],"(RAG)":[4],"for":[5],"complex":[6,202],"document":[7,203],"understanding":[8],"presents":[9],"a":[10,32,37,90,94,133,159],"formidable":[11],"dual":[12],"dilemma":[13],"of":[14],"accuracy":[15],"and":[16,86,118,175,198,221],"efficiency,":[17],"particularly":[18],"in":[19],"graph":[20,194],"RAG.":[21],"Processing":[22],"structurally":[23],"sparse":[24,67],"yet":[25],"visually":[26],"dense":[27],"layouts,":[28],"such":[29],"as":[30],"extracting":[31],"tiny":[33],"data":[34],"marker":[35],"from":[36,142,183],"financial":[38],"chart,":[39],"often":[40],"incurs":[41,103],"computationally":[42],"prohibitive":[43],"token":[44],"overhead":[45],"while":[46,100],"still":[47],"triggering":[48],"catastrophic":[49,97],"hallucination.":[50],"However,":[51],"Graph":[53,135],"RAG":[54,136],"pipelines":[55],"rely":[56],"on":[57],"graph-construction":[58,156,228],"stages":[59],"that":[60,77,110,119,138],"assume":[61],"Vision-Language":[62],"Models":[63],"(VLMs)":[64],"can":[65],"resolve":[66],"semantics":[68],"within":[69],"high-density":[70],"layouts.":[71],"We":[72],"challenge":[73],"this":[74,111,127],"assumption,":[75],"revealing":[76],"forcing":[78],"VLMs":[79],"to":[80,146,163,187,213,223],"localize":[81],"visual":[82],"evidence,":[83],"interpret":[84],"semantics,":[85],"extract":[87],"relations":[88],"triggers":[89],"\"Visual":[91],"Attention":[92],"Sink,\"":[93],"mechanism":[95],"driving":[96],"semantic":[98,120,190,208],"loss,":[99],"full-page":[101],"processing":[102,141,152],"massive":[104],"computational":[105],"overhead.":[106],"Controlled":[107],"interventions":[108],"verify":[109],"failure":[112],"is":[113],"boundary-driven":[114],"rather":[115],"than":[116,150],"content-specific":[117],"anchoring":[121],"mitigates":[122],"it.":[123],"To":[124],"fundamentally":[125],"correct":[126],"flawed":[128],"paradigm,":[129],"we":[130],"introduce":[131],"DeCoRAG,":[132],"pipeline":[137],"shifts":[139],"knowledge":[140],"coupled":[143],"visual-semantic":[144],"reasoning":[145,181],"\"Cognitive":[147],"Decoupling.\"":[148],"Rather":[149],"passively":[151],"raw":[153],"pixels,":[154],"its":[155],"stage":[157],"establishes":[158],"macroscopic":[160],"Semantic":[161],"Anchor":[162],"neutralize":[164],"the":[165,180,207,218],"attention":[166],"sink.":[167],"This":[168],"anchor":[169],"subsequently":[170],"drives":[171],"our":[172],"Region-Aware":[173],"Pruning":[174],"Cropping":[176],"(RAP-Crop)":[177],"mechanism,":[178],"shifting":[179],"space":[182],"dense,":[184],"noisy":[185],"backgrounds":[186],"purified,":[188],"intent-driven":[189],"clusters.":[191],"The":[192],"resulting":[193],"supports":[195],"hybrid":[196],"retrieval":[197],"answer":[199],"generation.":[200],"Across":[201],"benchmarks,":[204],"DeCoRAG":[205],"improves":[206],"pass":[209],"rate":[210],"by":[211,231],"up":[212],"12.5":[214],"percentage":[215],"points":[216],"over":[217],"strongest":[219],"baseline":[220],"generalizes":[222],"DocVQA.":[224],"RAP-Crop":[225],"reduces":[226],"offline":[227],"prompt":[229],"tokens":[230],"40.8%":[232],"without":[233],"sacrificing":[234],"end-to-end":[235],"accuracy.":[236]},"counts_by_year":[],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2026-07-29T00:00:00"}