{"id":"https://openalex.org/W7204946010","doi":"https://doi.org/10.48550/arxiv.2608.30753","title":"Learning from What You Retrieve: Online RL Fine-Tuning for Semantic Retrieval","display_name":"Learning from What You Retrieve: Online RL Fine-Tuning for Semantic Retrieval","publication_year":2026,"publication_date":"2026-08-31","ids":{"openalex":"https://openalex.org/W7204946010","doi":"https://doi.org/10.48550/arxiv.2608.30753"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.30753","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30753","pdf_url":"https://arxiv.org/pdf/2608.30753","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","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.30753","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5034717888","display_name":"Shaowei Wei","orcid":"https://orcid.org/0000-0001-8632-6113","observed_orcids":["https://orcid.org/0000-0001-8632-6113"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei, Shaowei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001573772","display_name":"Chong Huang","orcid":"https://orcid.org/0000-0002-0392-9398","observed_orcids":["https://orcid.org/0000-0002-0392-9398","https://orcid.org/0000-0003-0779-8214","https://orcid.org/0000-0002-4312-0382","https://orcid.org/0009-0004-3693-3141"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Chong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022166793","display_name":"Songtao Fang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fang, Songtao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148835341","display_name":"Jin Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148925151","display_name":"Zhuojun Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zhuojun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5028822516","display_name":"Chengfu Huo","orcid":"https://orcid.org/0000-0003-3611-6921","observed_orcids":["https://orcid.org/0000-0003-3611-6921"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huo, Chengfu","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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9897000193595886,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9897000193595886,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10203","display_name":"Recommender Systems and Techniques","score":0.0052999998442828655,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.002899999963119626,"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/semantic-search","display_name":"semantic search","score":0.8600000143051147},{"id":"https://openalex.org/keywords/re-ranking","display_name":"re-ranking","score":0.5230000019073486},{"id":"https://openalex.org/keywords/embedding-geometry","display_name":"embedding geometry","score":0.4059999883174896},{"id":"https://openalex.org/keywords/reward-modeling","display_name":"reward modeling","score":0.4749999940395355},{"id":"https://openalex.org/keywords/e-commerce-search","display_name":"e-commerce search","score":0.6460000276565552},{"id":"https://openalex.org/keywords/pao","display_name":"PAO","score":0.3160000145435333}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7775999903678894},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.7434999942779541},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.6219000220298767},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6069999933242798},{"id":"https://openalex.org/C2779532271","wikidata":"https://www.wikidata.org/wiki/Q445558","display_name":"Relevance feedback","level":4,"score":0.5724999904632568},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5681999921798706},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.47269999980926514},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.40799999237060547},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.3774999976158142},{"id":"https://openalex.org/C2776321320","wikidata":"https://www.wikidata.org/wiki/Q857525","display_name":"Annotation","level":2,"score":0.36959999799728394},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.3483000099658966},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33820000290870667},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.29249998927116394},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.27810001373291016},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.27559998631477356},{"id":"https://openalex.org/C130318100","wikidata":"https://www.wikidata.org/wiki/Q2268914","display_name":"Semantic similarity","level":2,"score":0.26989999413490295},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.26930001378059387},{"id":"https://openalex.org/C1667742","wikidata":"https://www.wikidata.org/wiki/Q10927554","display_name":"Image retrieval","level":3,"score":0.26750001311302185},{"id":"https://openalex.org/C161156560","wikidata":"https://www.wikidata.org/wiki/Q1638872","display_name":"Document retrieval","level":2,"score":0.2619999945163727},{"id":"https://openalex.org/C99016210","wikidata":"https://www.wikidata.org/wiki/Q5488129","display_name":"Query expansion","level":2,"score":0.2556000053882599}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.30753","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30753","pdf_url":"https://arxiv.org/pdf/2608.30753","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.2608.30753","doi":"https://doi.org/10.48550/arxiv.2608.30753","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.30753","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":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.30753","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30753","pdf_url":"https://arxiv.org/pdf/2608.30753","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7220922298","https://openalex.org/W7203867589","https://openalex.org/W4389470748","https://openalex.org/W7138879855","https://openalex.org/W7106124804","https://openalex.org/W7163871828","https://openalex.org/W7167911953","https://openalex.org/W7164886935","https://openalex.org/W7157969354","https://openalex.org/W7127606770"],"abstract_inverted_index":{"In":[0],"large-scale":[1],"e-commerce":[2],"retrieval,":[3],"dual-encoder":[4],"retrievers":[5],"are":[6],"op-":[7],"timized":[8],"for":[9,36],"contrastive":[10],"similarity,":[11],"whereas":[12],"downstream":[13],"rerankers":[14],"capture":[15],"finer-grained":[16],"relevance":[17],"preferences;":[18],"this":[19],"objective":[20],"mis-":[21],"match":[22],"limits":[23],"end-to-end":[24],"retrieval":[25],"quality.":[26],"Reinforcement":[27],"Learning":[28],"offers":[29],"a":[30,69,87,125],"way":[31],"to":[32,59,102],"use":[33],"reward-model":[34],"feedback":[35],"retriever":[37],"adaptation,":[38],"but":[39],"we":[40,65],"observe":[41],"that":[42,77,133],"standard":[43,138],"policy-gradient":[44],"updates":[45,100],"can":[46],"degrade":[47],"embedding":[48],"geometry,":[49],"especially":[50],"when":[51],"the":[52],"document":[53],"index":[54],"must":[55],"remain":[56],"frozen":[57,88],"due":[58],"industrial":[60,127],"constraints.":[61],"To":[62],"address":[63],"this,":[64],"propose":[66],"PAO":[67,96,134],"(Positive-Advantage-Only),":[68],"selective":[70],"RL":[71,139],"optimization":[72],"method.":[73],"Our":[74],"analysis":[75],"reveals":[76],"in-":[78],"discriminate":[79],"penalization":[80],"of":[81],"negative":[82],"samples":[83],"(pushing":[84],"away)":[85],"in":[86],"high-dimensional":[89],"space":[90],"disrupts":[91],"pre-trained":[92],"semantic":[93],"man-":[94],"ifolds.":[95],"selectively":[97],"applies":[98],"gradient":[99],"only":[101],"retrieved":[103],"items":[104],"with":[105],"positive":[106],"advantages,":[107],"effectively":[108],"pulling":[109],"query":[110],"embed-":[111],"dings":[112],"toward":[113],"high-reward":[114],"regions":[115],"while":[116],"preserving":[117],"global":[118],"topo-":[119],"logical":[120],"stability.":[121],"Experiments":[122],"on":[123],"both":[124],"massive":[126],"dataset":[128],"and":[129,140],"public":[130],"benchmarks":[131],"demonstrate":[132],"significantly":[135],"outper-":[136],"forms":[137],"distillation":[141],"baselines.":[142]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}