{"id":"https://openalex.org/W7204088728","doi":"https://doi.org/10.48550/arxiv.2608.21165","title":"Distilling Black-Box Machine Learning into a Small, Self-Explaining Language Model for Learning Analytics","display_name":"Distilling Black-Box Machine Learning into a Small, Self-Explaining Language Model for Learning Analytics","publication_year":2026,"publication_date":"2026-08-21","ids":{"openalex":"https://openalex.org/W7204088728","doi":"https://doi.org/10.48550/arxiv.2608.21165"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2608.21165","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21165","pdf_url":"https://arxiv.org/pdf/2608.21165","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.21165","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5054395060","display_name":"Chenguang Pan","orcid":"https://orcid.org/0009-0002-6507-8952","observed_orcids":["https://orcid.org/0009-0002-6507-8952"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Pan, Chenguang","raw_affiliation_strings":["Department of Human Development , Teachers College Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Human Development , Teachers College Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148034569","display_name":"Airui Meng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Meng, Airui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5027938311","display_name":"Youmi Suk","orcid":"https://orcid.org/0000-0003-0316-6201","observed_orcids":["https://orcid.org/0000-0003-0316-6201"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Suk, Youmi","raw_affiliation_strings":["Department of Human Development , Teachers College Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Human Development , Teachers College Columbia University","institution_ids":["https://openalex.org/I78577930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.6435999870300293,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.6435999870300293,"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/T11122","display_name":"Online Learning and Analytics","score":0.35030001401901245,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.00039999998989515007,"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/knowledge-distillation","display_name":"knowledge distillation","score":0.9589999914169312},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8790000081062317},{"id":"https://openalex.org/keywords/learning-analytics","display_name":"learning analytics","score":0.7329999804496765},{"id":"https://openalex.org/keywords/model-interpretability","display_name":"model interpretability","score":0.6330000162124634},{"id":"https://openalex.org/keywords/post-hoc-explanation","display_name":"post-hoc explanation","score":0.5889999866485596},{"id":"https://openalex.org/keywords/two-stage-fine-tuning","display_name":"two-stage fine-tuning","score":0.8190000057220459},{"id":"https://openalex.org/keywords/faithfulness-evaluation","display_name":"faithfulness evaluation","score":0.6370000243186951},{"id":"https://openalex.org/keywords/attribution","display_name":"attribution","score":0.35199999809265137},{"id":"https://openalex.org/keywords/personalized-prediction","display_name":"personalized prediction","score":0.6140000224113464},{"id":"https://openalex.org/keywords/model-opacity","display_name":"model opacity","score":0.43700000643730164},{"id":"https://openalex.org/keywords/educational-decision-making","display_name":"educational decision-making","score":0.4309999942779541},{"id":"https://openalex.org/keywords/mathematics-coursework","display_name":"mathematics coursework","score":0.5669999718666077}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7669000029563904},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.7621999979019165},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6743000149726868},{"id":"https://openalex.org/C55166926","wikidata":"https://www.wikidata.org/wiki/Q2892946","display_name":"Oracle","level":2,"score":0.5525000095367432},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.4528000056743622},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.42730000615119934},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.41530001163482666},{"id":"https://openalex.org/C79158427","wikidata":"https://www.wikidata.org/wiki/Q485396","display_name":"Analytics","level":2,"score":0.41429999470710754},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.37540000677108765},{"id":"https://openalex.org/C199521495","wikidata":"https://www.wikidata.org/wiki/Q181487","display_name":"Audit","level":2,"score":0.37290000915527344},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.34369999170303345},{"id":"https://openalex.org/C97256817","wikidata":"https://www.wikidata.org/wiki/Q1462316","display_name":"Spurious relationship","level":2,"score":0.3418999910354614},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.3345000147819519},{"id":"https://openalex.org/C94124525","wikidata":"https://www.wikidata.org/wiki/Q912550","display_name":"Categorization","level":2,"score":0.3084999918937683},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2874999940395355},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.28439998626708984},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2831999957561493},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.26899999380111694},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.26759999990463257},{"id":"https://openalex.org/C2777413886","wikidata":"https://www.wikidata.org/wiki/Q3276013","display_name":"Fluency","level":2,"score":0.26649999618530273},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.26260000467300415},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.258899986743927},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C168065819","wikidata":"https://www.wikidata.org/wiki/Q845566","display_name":"Debugging","level":2,"score":0.2533000111579895}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.21165","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21165","pdf_url":"https://arxiv.org/pdf/2608.21165","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.21165","doi":"https://doi.org/10.48550/arxiv.2608.21165","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.21165","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":"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":"pmh:oai:arXiv.org:2608.21165","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21165","pdf_url":"https://arxiv.org/pdf/2608.21165","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[{"score":0.8278999924659729,"display_name":"Quality education","id":"https://metadata.un.org/sdg/4"}],"sustainable_development_goals_aurora":[{"score":0.7983654737472534,"display_name":"Quality education","id":"https://metadata.un.org/sdg/4"}],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204088728.pdf","grobid_xml":"https://content.openalex.org/works/W7204088728.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Learning":[0],"analytics":[1],"increasingly":[2],"relies":[3],"on":[4,227],"flexible":[5],"machine":[6],"learning":[7],"(ML),":[8],"but":[9],"the":[10,14,52,80,123,131,171,184,187,202,209,236],"model":[11,50,117],"opacity":[12],"and":[13,37,59,68,134,166,211,224],"burden":[15],"of":[16,156,163,206],"deployment":[17],"prevent":[18],"these":[19],"tools":[20],"from":[21,95],"reaching":[22],"educational":[23],"practice.":[24],"We":[25,86,149],"propose":[26],"a":[27,33,45,71,88,104,114,140,180,218,228],"two-stage":[28],"fine-tuning":[29],"pipeline":[30,185],"that":[31,54,75,91,151,189,222],"distills":[32],"fitted":[34],"black-box":[35],"estimator":[36,96],"its":[38],"post":[39],"hoc":[40],"interpretation":[41],"(the":[42],"mentor)":[43],"into":[44],"small,":[46],"open-weight":[47],"large":[48],"language":[49],"(LLM;":[51],"mentee)":[53],"returns":[55],"an":[56,100,109],"individual-level":[57],"estimate":[58],"explains":[60,225],"in":[61,121,174,199],"natural":[62],"language.":[63],"The":[64,215],"design":[65,87],"is":[66,118,153,161,217],"estimator-agnostic":[67],"paired":[69],"with":[70,103,113,204],"faithfulness-first":[72],"evaluation":[73],"framework":[74],"audits":[76],"every":[77],"narration":[78,159],"against":[79],"attribution":[81],"it":[82],"claims":[83],"to":[84,179,197],"describe.":[85],"simulation":[89],"study":[90],"separates":[92],"distillation":[93,112],"loss":[94,97],"by":[98],"comparing":[99],"oracle":[101,110],"mentor":[102],"realistic":[105,141],"ML":[106],"mentor.":[107],"Given":[108],"signal,":[111],"two-billion-parameter":[115],"LLM":[116,221],"nearly":[119],"lossless":[120],"recovering":[122],"effect":[124],"surface":[125],"(r":[126],">":[127],".90),":[128],"perfectly":[129],"ranking":[130],"important":[132],"variables,":[133],"citing":[135],"no":[136,154,212],"spurious":[137],"covariate.":[138],"Under":[139],"estimator,":[142],"almost":[143],"all":[144],"remaining":[145],"error":[146],"originates":[147],"upstream.":[148],"find":[150],"fluency":[152],"evidence":[155],"correctness":[157],"since":[158],"quality":[160,168],"independent":[162],"signal":[164],"quality,":[165],"decision":[167],"collapses":[169],"toward":[170],"majority":[172],"action":[173],"severely":[175],"imbalanced":[176],"settings.":[177],"Applied":[178],"nationally":[181],"representative":[182],"dataset,":[183],"recovers":[186],"finding":[188],"advanced":[190],"mathematics":[191],"coursework":[192],"benefits":[193],"students":[194],"least":[195],"likely":[196],"enroll":[198],"four-year":[200],"college":[201],"most,":[203],"98.8%":[205],"narrations":[207],"passing":[208],"audit":[210],"fabricated":[213],"quantities.":[214],"result":[216],"single":[219],"fine-tuned":[220],"predicts":[223],"offline":[226],"commodity":[229],"laptop,":[230],"so":[231],"student":[232],"records":[233],"never":[234],"leave":[235],"machine.":[237]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2026-08-25T00:00:00"}