{"id":"https://openalex.org/W7220650587","doi":"https://doi.org/10.48550/arxiv.2610.06603","title":"Word-Level Text Unmixing via Evidence-Preserving Ownership Routing with Language Models","display_name":"Word-Level Text Unmixing via Evidence-Preserving Ownership Routing with Language Models","publication_year":2026,"publication_date":"2026-10-05","ids":{"openalex":"https://openalex.org/W7220650587","doi":"https://doi.org/10.48550/arxiv.2610.06603"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.06603","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.06603","pdf_url":"https://arxiv.org/pdf/2610.06603","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2610.06603","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114096762","display_name":"Jinglin He","orcid":"https://orcid.org/0009-0007-4221-5447","observed_orcids":["https://orcid.org/0009-0007-4221-5447"]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"He, Jinglin","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157772763","display_name":"Siyang Jiang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Jiang, Siyang","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157765020","display_name":"Lixing He","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"He, Lixing","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072780737","display_name":"Guoliang Xing","orcid":"https://orcid.org/0000-0003-1772-7751","observed_orcids":["https://orcid.org/0000-0003-1772-7751"]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xing, Guoliang","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5157675703","display_name":"Hongkai Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Chen, Hongkai","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I177725633"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T12380","display_name":"Authorship Attribution and Profiling","score":0.45680001378059387,"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/T12380","display_name":"Authorship Attribution and Profiling","score":0.45680001378059387,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.29269999265670776,"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/T10028","display_name":"Topic Modeling","score":0.15330000221729279,"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/causal-language-models","display_name":"causal language models","score":0.7630000114440918},{"id":"https://openalex.org/keywords/constrained-decoding","display_name":"constrained decoding","score":0.753000020980835},{"id":"https://openalex.org/keywords/deterministic-reconstruction","display_name":"deterministic reconstruction","score":0.6570000052452087},{"id":"https://openalex.org/keywords/speech-transcripts","display_name":"speech transcripts","score":0.6349999904632568}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7925000190734863},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.5687999725341797},{"id":"https://openalex.org/C108757681","wikidata":"https://www.wikidata.org/wiki/Q2773912","display_name":"Bigram","level":3,"score":0.539900004863739},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5210999846458435},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.4968000054359436},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.4767000079154968},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4713999927043915},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4465000033378601},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.42010000348091125},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.39899998903274536},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.37540000677108765},{"id":"https://openalex.org/C554936623","wikidata":"https://www.wikidata.org/wiki/Q199657","display_name":"Reading (process)","level":2,"score":0.3686999976634979},{"id":"https://openalex.org/C93518851","wikidata":"https://www.wikidata.org/wiki/Q180160","display_name":"Metadata","level":2,"score":0.34380000829696655},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.33709999918937683},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.3179999887943268},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.3158999979496002},{"id":"https://openalex.org/C159254197","wikidata":"https://www.wikidata.org/wiki/Q1144915","display_name":"Lexicographical order","level":2,"score":0.31189998984336853},{"id":"https://openalex.org/C2780922921","wikidata":"https://www.wikidata.org/wiki/Q255189","display_name":"Paraphrase","level":2,"score":0.30720001459121704},{"id":"https://openalex.org/C36299963","wikidata":"https://www.wikidata.org/wiki/Q1369844","display_name":"Observability","level":2,"score":0.3034000098705292},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.30230000615119934},{"id":"https://openalex.org/C70777604","wikidata":"https://www.wikidata.org/wiki/Q257885","display_name":"Word order","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C2779439875","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Natural language understanding","level":3,"score":0.28529998660087585},{"id":"https://openalex.org/C141513077","wikidata":"https://www.wikidata.org/wiki/Q378542","display_name":"Independent and identically distributed random variables","level":3,"score":0.28279998898506165},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2766000032424927},{"id":"https://openalex.org/C154690210","wikidata":"https://www.wikidata.org/wiki/Q1668499","display_name":"Rewriting","level":2,"score":0.2752000093460083},{"id":"https://openalex.org/C146834321","wikidata":"https://www.wikidata.org/wiki/Q2979672","display_name":"Closure (psychology)","level":2,"score":0.2745000123977661},{"id":"https://openalex.org/C126706616","wikidata":"https://www.wikidata.org/wiki/Q2944660","display_name":"Lexical item","level":2,"score":0.2709999978542328},{"id":"https://openalex.org/C75778745","wikidata":"https://www.wikidata.org/wiki/Q342626","display_name":"Lag","level":2,"score":0.2678999900817871},{"id":"https://openalex.org/C2778484313","wikidata":"https://www.wikidata.org/wiki/Q1172540","display_name":"Data stream","level":2,"score":0.26579999923706055},{"id":"https://openalex.org/C204983608","wikidata":"https://www.wikidata.org/wiki/Q2111958","display_name":"Productivity","level":2,"score":0.2630999982357025},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C191399111","wikidata":"https://www.wikidata.org/wiki/Q64861","display_name":"Transitive relation","level":2,"score":0.25619998574256897},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.2533000111579895}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.06603","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.06603","pdf_url":"https://arxiv.org/pdf/2610.06603","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.06603","doi":"https://doi.org/10.48550/arxiv.2610.06603","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.06603","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,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2610.06603","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.06603","pdf_url":"https://arxiv.org/pdf/2610.06603","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220650587.pdf","grobid_xml":"https://content.openalex.org/works/W7220650587.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7221822297","https://openalex.org/W339218405","https://openalex.org/W6885143814","https://openalex.org/W6992439537","https://openalex.org/W4385571740","https://openalex.org/W7134963394","https://openalex.org/W7172450107","https://openalex.org/W4223480327","https://openalex.org/W7172506806","https://openalex.org/W4412887709"],"abstract_inverted_index":{"Text":[0,34],"from":[1,87,141,148,205],"multiple":[2],"sources":[3],"can":[4,66],"become":[5],"interleaved":[6,38],"into":[7],"a":[8,91,159,209],"single":[9],"sequence":[10],"when":[11,196],"attribution":[12],"metadata":[13],"is":[14,113,199],"lost,":[15],"such":[16],"as":[17,32],"overlapping":[18],"speech":[19,140],"transcripts,":[20],"document":[21,146],"reading":[22],"flows,":[23],"or":[24,69],"concurrent":[25,152],"agent":[26],"streams.":[27],"We":[28,76,131],"formalize":[29],"this":[30,73],"challenge":[31],"Word-Level":[33],"Unmixing:":[35],"given":[36],"an":[37],"lexical":[39,197,206],"stream":[40,103],"and":[41,55,104,143,150,185],"source":[42,48],"count":[43],"K,":[44],"recover":[45],"the":[46,101,164,175],"original":[47],"sequences":[49],"while":[50],"preserving":[51],"every":[52,126],"word":[53,168],"occurrence":[54,128],"its":[56],"within-source":[57],"order":[58],"exactly.":[59],"Directly":[60],"generating":[61],"separated":[62],"texts":[63],"with":[64,115,188],"LLMs":[65],"omit,":[67],"duplicate,":[68],"hallucinate":[70],"words,":[71],"violating":[72],"exact-reconstruction":[74],"objective.":[75],"therefore":[77],"propose":[78],"Evidence-Preserving":[79],"Ownership":[80],"Routing":[81],"(EPOR),":[82],"which":[83],"decouples":[84],"source-ownership":[85],"prediction":[86],"reconstruction.":[88],"EPOR":[89,161],"adapts":[90],"causal":[92],"LLM":[93],"to":[94,181,212],"predict":[95],"canonical":[96],"ownership":[97,203],"routes":[98],"conditioned":[99],"on":[100],"mixed":[102],"prior":[105],"routing":[106],"decisions.":[107],"At":[108],"inference,":[109],"completion-safe":[110],"constrained":[111],"decoding":[112],"combined":[114],"deterministic":[116],"indexed":[117],"reconstruction,":[118],"yielding":[119],"structurally":[120],"valid":[121],"K-source":[122],"partitions":[123],"that":[124,195],"preserve":[125],"observed":[127],"exactly":[129],"once.":[130],"also":[132],"introduce":[133],"UNMIXBENCH,":[134],"covering":[135],"controlled":[136],"synthetic":[137],"mixtures,":[138],"timestamp-derived":[139],"AMI":[142],"ICSI,":[144],"layout-derived":[145],"streams":[147],"ReadingBank,":[149],"simulated":[151],"digital":[153],"outputs.":[154],"Across":[155],"five":[156],"evaluation":[157],"tracks,":[158],"4B":[160],"model":[162],"achieves":[163],"lowest":[165],"mean":[166,177],"minimum-permutation":[167],"error":[169],"rate":[170],"among":[171],"finetuned":[172],"baselines,":[173],"reducing":[174],"five-track":[176],"by":[178],"22.3%":[179],"relative":[180],"compact":[182],"source-array":[183],"generation":[184],"remaining":[186],"competitive":[187],"zero-shot":[189],"frontier":[190],"LLMs.":[191],"These":[192],"results":[193],"show":[194],"evidence":[198],"fully":[200],"observed,":[201],"separating":[202],"inference":[204],"regeneration":[207],"provides":[208],"reliable":[210],"alternative":[211],"direct":[213],"generation.":[214]},"counts_by_year":[],"updated_date":"2026-11-06T12:05:46.655746","created_date":"2026-10-07T00:00:00"}