{"id":"https://openalex.org/W7167695864","doi":"https://doi.org/10.1109/access.2026.3711096","title":"Beyond VQA Accuracy: A Cross-Regime Diagnostic Evaluation of Symbol Consistency in Multimodal Large Language Models","display_name":"Beyond VQA Accuracy: A Cross-Regime Diagnostic Evaluation of Symbol Consistency in Multimodal Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7167695864","doi":"https://doi.org/10.1109/access.2026.3711096"},"language":"en","primary_location":{"id":"doi:10.1109/access.2026.3711096","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2026.3711096","pdf_url":null,"source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1109/access.2026.3711096","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5012605939","display_name":"Chun Wang","orcid":"https://orcid.org/0000-0001-9910-5483"},"institutions":[{"id":"https://openalex.org/I4210120842","display_name":"Shenyang Institute of Computing Technology (China)","ror":"https://ror.org/02xqp7x96","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210120842"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chuanchen Wang","raw_affiliation_strings":["Shenyang Institute of Computing Technology"],"raw_orcid":"https://orcid.org/0009-0005-8149-558X","affiliations":[{"raw_affiliation_string":"Shenyang Institute of Computing Technology","institution_ids":["https://openalex.org/I4210120842"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019655330","display_name":"Ruifeng Guo","orcid":"https://orcid.org/0009-0000-9075-1884"},"institutions":[{"id":"https://openalex.org/I4210120842","display_name":"Shenyang Institute of Computing Technology (China)","ror":"https://ror.org/02xqp7x96","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210120842"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ruifeng Guo","raw_affiliation_strings":["Shenyang Institute of Computing Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shenyang Institute of Computing Technology","institution_ids":["https://openalex.org/I4210120842"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007887582","display_name":"Pengfei Xiu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210120842","display_name":"Shenyang Institute of Computing Technology (China)","ror":"https://ror.org/02xqp7x96","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210120842"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Pengfei Xiu","raw_affiliation_strings":["Shenyang Institute of Computing Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shenyang Institute of Computing Technology","institution_ids":["https://openalex.org/I4210120842"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008824186","display_name":"Yu Zhao","orcid":"https://orcid.org/0000-0003-1874-908X"},"institutions":[{"id":"https://openalex.org/I4210120842","display_name":"Shenyang Institute of Computing Technology (China)","ror":"https://ror.org/02xqp7x96","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210120842"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yanqing Zhao","raw_affiliation_strings":["Shenyang Institute of Computing Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shenyang Institute of Computing Technology","institution_ids":["https://openalex.org/I4210120842"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100412686","display_name":"Hongliang Wang","orcid":"https://orcid.org/0000-0002-5865-8898"},"institutions":[{"id":"https://openalex.org/I4210120842","display_name":"Shenyang Institute of Computing Technology (China)","ror":"https://ror.org/02xqp7x96","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210120842"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongliang Wang","raw_affiliation_strings":["Shenyang Institute of Computing Technology"],"raw_orcid":"https://orcid.org/0009-0007-5484-286X","affiliations":[{"raw_affiliation_string":"Shenyang Institute of Computing Technology","institution_ids":["https://openalex.org/I4210120842"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210120842"],"apc_list":{"value":1850,"currency":"USD","value_usd":1850},"apc_paid":{"value":1850,"currency":"USD","value_usd":1850},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.86335987,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"14","issue":null,"first_page":"109309","last_page":"109331"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.2540999948978424,"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"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.2540999948978424,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.20659999549388885,"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/T13629","display_name":"Text Readability and Simplification","score":0.12800000607967377,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.6657999753952026},{"id":"https://openalex.org/keywords/symbol","display_name":"Symbol (formal)","score":0.6485999822616577},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.3100000023841858},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.29120001196861267},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.28790000081062317},{"id":"https://openalex.org/keywords/strong-consistency","display_name":"Strong consistency","score":0.275299996137619}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7800999879837036},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6657999753952026},{"id":"https://openalex.org/C134400042","wikidata":"https://www.wikidata.org/wiki/Q2372244","display_name":"Symbol (formal)","level":2,"score":0.6485999822616577},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5874000191688538},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5789999961853027},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.3100000023841858},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.29120001196861267},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.28790000081062317},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2831000089645386},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.2766000032424927},{"id":"https://openalex.org/C31388003","wikidata":"https://www.wikidata.org/wiki/Q7624548","display_name":"Strong consistency","level":3,"score":0.275299996137619},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.27149999141693115},{"id":"https://openalex.org/C93361087","wikidata":"https://www.wikidata.org/wiki/Q4426698","display_name":"Data consistency","level":2,"score":0.26350000500679016},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2581000030040741},{"id":"https://openalex.org/C3020028006","wikidata":"https://www.wikidata.org/wiki/Q9158","display_name":"Electronic mail","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/access.2026.3711096","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2026.3711096","pdf_url":null,"source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:1e9dc19a8a98436b935db1d2bc48f027","is_oa":false,"landing_page_url":"https://doaj.org/article/1e9dc19a8a98436b935db1d2bc48f027","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"IEEE Access, Vol 14, Pp 109309-109331 (2026)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.1109/access.2026.3711096","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2026.3711096","pdf_url":null,"source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"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":{"Multimodal":[0],"large":[1],"language":[2],"models":[3],"(MLLMs)":[4],"have":[5],"achieved":[6],"strong":[7],"performance":[8],"in":[9,18,74,219,253],"general":[10,174,233],"visual":[11,16,141],"question":[12],"answering,":[13],"yet":[14],"their":[15],"faithfulness":[17],"handling":[19],"discrete":[20],"symbolic":[21],"information":[22],"remains":[23],"insufficiently":[24],"understood.":[25],"Symbols":[26],"such":[27],"as":[28,147],"numbers,":[29],"time":[30],"expressions,":[31],"identifiers,":[32],"license":[33],"plates,":[34],"and":[35,42,90,106,158,166,207,224,236,250],"alphanumeric":[36],"strings":[37],"impose":[38],"low":[39],"semantic":[40],"redundancy":[41],"strict":[43],"character-level":[44,220],"constraints,":[45],"making":[46],"overall":[47],"VQA":[48,175,234],"accuracy":[49,160,176],"or":[50],"isolated":[51],"OCR-style":[52],"evaluation":[53,235],"inadequate":[54],"for":[55,70,183,247],"diagnosing":[56],"model":[57,168],"reliability.":[58],"To":[59],"address":[60],"this":[61,63],"gap,":[62],"paper":[64],"proposes":[65],"a":[66,77,95,116,154,243],"cross-regime":[67],"diagnostic":[68,245],"framework":[69],"evaluating":[71],"symbol":[72,104,132,230],"consistency":[73,187,231],"MLLMs.":[75,254],"Under":[76],"unified":[78],"protocol,":[79],"we":[80],"evaluate":[81],"five":[82],"representative":[83],"models,":[84,215],"including":[85],"BLIP-2,":[86],"InstructBLIP,":[87],"LLaVA,":[88],"InternVL,":[89],"Qwen,":[91],"across":[92],"VQA,":[93],"TextVQA,":[94],"self-constructed":[96],"Symbol":[97,186],"Subset,":[98],"Regime":[99,107,143],"2a":[100,144],"with":[101,109],"uncontrolled":[102,149],"generative":[103],"rendering,":[105],"2b":[108],"controlled":[110],"clear-symbol":[111,211],"grounding.":[112],"We":[113],"further":[114],"introduce":[115],"2a<inline-formula>":[117],"<tex-math":[118],"notation=\"LaTeX\">$\\rightarrow":[119],"2$":[120],"</tex-math></inline-formula>b":[121],"paired":[122],"recovery":[123],"analysis":[124],"to":[125],"distinguish":[126],"rendering-sensitive":[127],"errors":[128,136],"caused":[129],"by":[130],"upstream":[131,163],"degradation":[133],"from":[134,197,232],"persistent":[135,216],"that":[137,173],"remain":[138,218],"under":[139],"clear":[140],"evidence.":[142],"is":[145],"interpreted":[146],"an":[148],"stress":[150],"probe":[151],"rather":[152],"than":[153],"clean":[155],"OCR":[156,192],"benchmark,":[157],"its":[159,238],"reflects":[161],"both":[162],"rendering":[164],"quality":[165],"downstream":[167],"reading.":[169],"Experimental":[170],"results":[171],"show":[172],"can":[177],"substantially":[178],"overestimate":[179],"symbol-centered":[180],"reliability,":[181],"especially":[182],"weaker":[184],"models.":[185],"failures":[188,217],"are":[189],"not":[190],"merely":[191],"recognition":[193],"errors,":[194],"but":[195],"arise":[196],"the":[198],"interaction":[199],"of":[200],"target-region":[201],"binding,":[202,223],"character-faithful":[203],"transcription,":[204],"answer":[205],"completeness,":[206],"language-prior":[208],"normalization.":[209],"Although":[210],"conditions":[212],"improve":[213],"stronger":[214],"grounding,":[221],"target":[222],"task":[225],"following.":[226],"This":[227],"study":[228],"separates":[229],"reveals":[237],"multi-stage":[239],"failure":[240],"mechanisms,":[241],"offering":[242],"reusable":[244],"perspective":[246],"reliability":[248],"assessment":[249],"symbol-capability":[251],"improvement":[252]},"counts_by_year":[],"updated_date":"2026-07-25T09:21:30.201066","created_date":"2026-07-09T00:00:00"}