{"id":"https://openalex.org/W7170035278","doi":"https://doi.org/10.48550/arxiv.2607.19219","title":"Beyond Score Prediction: LLM-Based Essay Scoring and Feedback Generation via Reinforcement Learning with Rubric Rewards","display_name":"Beyond Score Prediction: LLM-Based Essay Scoring and Feedback Generation via Reinforcement Learning with Rubric Rewards","publication_year":2026,"publication_date":"2026-07-21","ids":{"openalex":"https://openalex.org/W7170035278","doi":"https://doi.org/10.48550/arxiv.2607.19219"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.19219","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.19219","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.2607.19219","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5040131467","display_name":"Xuefeng Jin","orcid":"https://orcid.org/0000-0002-4602-3107"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Xuefeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143442984","display_name":"Jiashuo Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiashuo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135952558","display_name":"Teng Cao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cao, Teng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5143454794","display_name":"Bin Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Bin","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/T10028","display_name":"Topic Modeling","score":0.40720000863075256,"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":0.40720000863075256,"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/T10664","display_name":"Sentiment Analysis and Opinion Mining","score":0.2484000027179718,"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/T11550","display_name":"Text and Document Classification Technologies","score":0.04170000180602074,"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/rubric","display_name":"Rubric","score":0.8001999855041504},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.6651999950408936},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.5623000264167786},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.5354999899864197},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.4505000114440918},{"id":"https://openalex.org/keywords/usable","display_name":"USable","score":0.42010000348091125},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.41029998660087585},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.40220001339912415}],"concepts":[{"id":"https://openalex.org/C111640148","wikidata":"https://www.wikidata.org/wiki/Q847349","display_name":"Rubric","level":2,"score":0.8001999855041504},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7516000270843506},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.6651999950408936},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.6276999711990356},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5623000264167786},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5612000226974487},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.5354999899864197},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.4505000114440918},{"id":"https://openalex.org/C2780615836","wikidata":"https://www.wikidata.org/wiki/Q2471869","display_name":"USable","level":2,"score":0.42010000348091125},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.41029998660087585},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.40220001339912415},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.35339999198913574},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.322299987077713},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.3142000138759613},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3084999918937683},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.27709999680519104},{"id":"https://openalex.org/C2779305910","wikidata":"https://www.wikidata.org/wiki/Q5172809","display_name":"Corrective feedback","level":2,"score":0.27619999647140503},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.275299996137619},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.2752000093460083},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.26429998874664307},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.25999999046325684},{"id":"https://openalex.org/C148524875","wikidata":"https://www.wikidata.org/wiki/Q6975395","display_name":"F1 score","level":2,"score":0.2596000134944916}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.19219","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.19219","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.2607.19219","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.19219","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":[{"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10","score":0.6628965139389038}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"have":[4],"been":[5],"widely":[6],"applied":[7],"to":[8,125,178],"automated":[9,14,38],"essay":[10,55],"scoring":[11,56,165],"(AES)":[12],"and":[13,37,57,68,89,151,180,190],"feedback":[15,41,58,64,80,105,116,175,183],"generation":[16,59],"(AFG).":[17],"However,":[18],"existing":[19],"studies":[20],"rely":[21],"primarily":[22],"on":[23,32,93,107],"prompt":[24],"engineering":[25],"or":[26],"supervised":[27],"fine-tuning,":[28],"while":[29,114,173],"systematic":[30],"research":[31],"reinforcement":[33],"learning":[34],"(RL)":[35],"post-training":[36],"evaluation":[39,81,112],"of":[40],"quality":[42,65,176],"remains":[43],"limited.":[44],"We":[45,118],"propose":[46,96,120],"RLAES,":[47],"a":[48],"unified":[49],"LLM":[50],"framework":[51,82,142],"that":[52,139],"jointly":[53],"optimizes":[54],"through":[60],"RL.":[61,188],"To":[62],"make":[63],"measurable,":[66],"interpretable,":[67],"usable":[69],"for":[70],"training,":[71],"we":[72,95],"introduce":[73],"Rubric-based":[74],"Feedback":[75,99],"Evaluation":[76],"(RFE),":[77],"an":[78,90],"essay-grounded":[79],"comprising":[83],"166":[84],"fine-grained":[85],"binary":[86],"rubric":[87],"items":[88],"LLM-as-judge.":[91],"Building":[92],"RFE,":[94],"Adaptive":[97],"Gated":[98],"Optimization":[100],"(AGFO),":[101],"which":[102],"activates":[103],"rubric-based":[104],"rewards":[106],"demand":[108],"during":[109],"RL,":[110],"reducing":[111],"overhead":[113],"improving":[115],"quality.":[117],"also":[119],"Adjacent":[121],"Contrastive":[122],"Reasoning":[123],"(ACR)":[124],"improve":[126],"ordinal":[127],"score":[128,134],"calibration":[129],"by":[130],"explicitly":[131],"contrasting":[132],"adjacent":[133],"levels.":[135],"Experimental":[136],"results":[137],"show":[138],"the":[140,158,163,182],"RFE":[141],"captures":[143],"essay-feedback":[144],"consistency,":[145],"exhibits":[146],"strong":[147],"pairwise":[148],"discriminative":[149],"power,":[150],"closely":[152],"aligns":[153],"with":[154],"expert":[155],"preferences.":[156],"On":[157],"ASAP":[159],"benchmark,":[160],"RLAES-AGFO":[161],"achieves":[162],"best":[164],"performance":[166],"among":[167],"LLM-based":[168],"methods":[169],"(QWK":[170],"=":[171],"0.803),":[172],"maintaining":[174],"comparable":[177],"GPT-5.5":[179],"avoiding":[181],"degradation":[184],"observed":[185],"under":[186],"score-only":[187],"Code":[189],"datasets":[191],"are":[192],"publicly":[193],"available":[194],"at":[195],"https://github.com/hellomuyi/RLAES.":[196]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-23T00:00:00"}