{"id":"https://openalex.org/W7165867757","doi":"https://doi.org/10.48550/arxiv.2606.25343","title":"Invoice Haystack: Benchmarking Document Retrieval and Visual Question Answering Under Strong Visual Homogeneity","display_name":"Invoice Haystack: Benchmarking Document Retrieval and Visual Question Answering Under Strong Visual Homogeneity","publication_year":2026,"publication_date":"2026-06-24","ids":{"openalex":"https://openalex.org/W7165867757","doi":"https://doi.org/10.48550/arxiv.2606.25343"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.25343","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25343","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":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.2606.25343","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114402838","display_name":"Heethanjan Kanagalingam","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kanagalingam, Heethanjan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114402839","display_name":"Thenukan Pathmanathan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pathmanathan, Thenukan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114402841","display_name":"Mokeeshan Vathanakumar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vathanakumar, Mokeeshan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139338262","display_name":"Basim Azam","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Azam, Basim","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070030398","display_name":"Sarah Erfani","orcid":"https://orcid.org/0000-0003-0885-0643"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Erfani, Sarah Monazam","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Akhtar, Naveed","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Akhtar, Naveed","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":0.9621999859809875,"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":0.9621999859809875,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.026799999177455902,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.0017000000225380063,"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/question-answering","display_name":"Question answering","score":0.6929000020027161},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.5131999850273132},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.46790000796318054},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.46560001373291016},{"id":"https://openalex.org/keywords/invoice","display_name":"Invoice","score":0.4641000032424927},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.45750001072883606},{"id":"https://openalex.org/keywords/document-retrieval","display_name":"Document retrieval","score":0.44339999556541443},{"id":"https://openalex.org/keywords/search-engine-indexing","display_name":"Search engine indexing","score":0.4099999964237213},{"id":"https://openalex.org/keywords/filter","display_name":"Filter (signal processing)","score":0.40560001134872437}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7282000184059143},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.6929000020027161},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.6414999961853027},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.5131999850273132},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.46790000796318054},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.46560001373291016},{"id":"https://openalex.org/C2781456945","wikidata":"https://www.wikidata.org/wiki/Q190581","display_name":"Invoice","level":2,"score":0.4641000032424927},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.45750001072883606},{"id":"https://openalex.org/C161156560","wikidata":"https://www.wikidata.org/wiki/Q1638872","display_name":"Document retrieval","level":2,"score":0.44339999556541443},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43540000915527344},{"id":"https://openalex.org/C75165309","wikidata":"https://www.wikidata.org/wiki/Q2258979","display_name":"Search engine indexing","level":2,"score":0.4099999964237213},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.40560001134872437},{"id":"https://openalex.org/C2780762811","wikidata":"https://www.wikidata.org/wiki/Q1784941","display_name":"Cosine similarity","level":3,"score":0.39079999923706055},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.36090001463890076},{"id":"https://openalex.org/C66882249","wikidata":"https://www.wikidata.org/wiki/Q169336","display_name":"Homogeneous","level":2,"score":0.3531000018119812},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.3456999957561493},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.335099995136261},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.334199994802475},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.33340001106262207},{"id":"https://openalex.org/C44083865","wikidata":"https://www.wikidata.org/wiki/Q3853443","display_name":"Mean reciprocal rank","level":2,"score":0.33250001072883606},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.30250000953674316},{"id":"https://openalex.org/C183115368","wikidata":"https://www.wikidata.org/wiki/Q856577","display_name":"Weighting","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.29179999232292175},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.28630000352859497},{"id":"https://openalex.org/C55166926","wikidata":"https://www.wikidata.org/wiki/Q2892946","display_name":"Oracle","level":2,"score":0.2799000144004822},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2793999910354614},{"id":"https://openalex.org/C1667742","wikidata":"https://www.wikidata.org/wiki/Q10927554","display_name":"Image retrieval","level":3,"score":0.2680000066757202},{"id":"https://openalex.org/C189391414","wikidata":"https://www.wikidata.org/wiki/Q7936579","display_name":"Visual Word","level":4,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.25343","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25343","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":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.2606.25343","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25343","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10","score":0.7236908674240112}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Language":[1],"Models":[2],"have":[3],"achieved":[4],"near-human":[5],"performance":[6],"on":[7,155,175],"single-document":[8],"Visual":[9],"Question":[10],"Answering,":[11],"yet":[12],"their":[13],"effectiveness":[14],"degrades":[15],"significantly":[16],"when":[17],"retrieving":[18],"information":[19],"from":[20],"large":[21],"collections":[22],"of":[23,49,95,138],"visually":[24],"homogeneous":[25,198],"documents.":[26],"Existing":[27],"multi-document":[28],"benchmarks":[29],"aggregate":[30],"diverse":[31],"document":[32,45,149,201],"types,":[33],"creating":[34],"artificial":[35],"separation":[36],"in":[37,104],"embedding":[38,59],"space":[39],"that":[40,126],"does":[41],"not":[42],"reflect":[43],"enterprise":[44],"repositories":[46],"where":[47],"thousands":[48],"records":[50],"share":[51],"identical":[52],"visual":[53,85,131],"templates.":[54],"We":[55],"identify":[56],"this":[57],"as":[58,190],"collapse":[60],"and":[61,101,130,180,199],"introduce":[62],"Invoice":[63,87,156],"Haystack,":[64],"a":[65,90,108,121,143,191],"benchmark":[66],"with":[67,73],"1,500":[68],"anonymized":[69],"invoice":[70],"images":[71],"paired":[72],"200":[74],"discriminative":[75],"question-answer":[76],"pairs,":[77],"specifically":[78],"designed":[79],"to":[80,98,133,164],"stress-test":[81],"retrieval":[82,112,173,194],"under":[83],"strong":[84],"homogeneity.":[86],"Haystack":[88],"exhibits":[89],"mean":[91],"pairwise":[92],"cosine":[93],"similarity":[94],"0.73,":[96],"compared":[97],"0.38":[99],"(DocHaystack)":[100],"0.31":[102],"(InfoHaystack)":[103],"existing":[105,159],"benchmarks,":[106],"posing":[107],"fundamentally":[109],"more":[110],"challenging":[111],"problem.":[113],"Addressing":[114],"the":[115,135,186],"identified":[116],"challenge,":[117],"we":[118],"propose":[119],"VL-RAG,":[120],"hybrid":[122],"retrieval-augmented":[123],"generation":[124],"framework":[125],"jointly":[127],"leverages":[128],"text":[129],"embeddings":[132],"harness":[134],"complementary":[136],"strengths":[137],"both":[139,197],"modalities,":[140],"followed":[141],"by":[142,162],"VLM-based":[144],"verification":[145],"filter":[146],"for":[147],"precise":[148],"identification.":[150],"VL-RAG":[151],"achieves":[152],"60.0\\%":[153],"Recall@1":[154],"Haystack-500,":[157],"outperforming":[158],"state-of-the-art":[160],"method":[161],"up":[163],"an":[165],"absolute":[166],"13.5":[167],"percentage":[168],"points.":[169],"It":[170],"further":[171],"improves":[172],"considerably":[174],"DocHaystack-1000":[176],"(77.1\\%":[177],"vs.\\":[178,183],"75.2\\%)":[179],"InfoHaystack-1000":[181],"(84.5\\%":[182],"80.0\\%),":[184],"establishing":[185],"proposed":[187],"dual-stream":[188],"fusion":[189],"consistently":[192],"superior":[193],"strategy":[195],"across":[196],"heterogeneous":[200],"collections.":[202]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-26T00:00:00"}