{"id":"https://openalex.org/W7168359331","doi":"https://doi.org/10.1145/3805712.3808444","title":"STAR: Staged Training with Aligned Reinforcement Learning and Multi-Faceted Distillation for Interpretable E-commerce Relevance","display_name":"STAR: Staged Training with Aligned Reinforcement Learning and Multi-Faceted Distillation for Interpretable E-commerce Relevance","publication_year":2026,"publication_date":"2026-07-15","ids":{"openalex":"https://openalex.org/W7168359331","doi":"https://doi.org/10.1145/3805712.3808444"},"language":"en","primary_location":{"id":"doi:10.1145/3805712.3808444","doi":"https://doi.org/10.1145/3805712.3808444","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805712.3808444","pdf_url":null,"source":{"id":"https://openalex.org/S4363608773","display_name":"International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1145/3805712.3808444","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5140955636","display_name":"Chenxu Wang","orcid":"https://orcid.org/0000-0001-5234-1454","observed_orcids":["https://orcid.org/0000-0001-5234-1454"]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chenxu Wang","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-5234-1454","affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008200779","display_name":"Jianzhi Shao","orcid":"https://orcid.org/0000-0002-1179-004X","observed_orcids":["https://orcid.org/0000-0002-1179-004X"]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jianzhi Shao","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-1179-004X","affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140792157","display_name":"Chi Zhang","orcid":"https://orcid.org/0009-0001-7624-439X","observed_orcids":["https://orcid.org/0009-0001-7624-439X"]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chi Zhang","raw_affiliation_strings":["Alibaba Group, Hangzhou, China"],"raw_orcid":"https://orcid.org/0009-0001-7624-439X","affiliations":[{"raw_affiliation_string":"Alibaba Group, Hangzhou, China","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5140959215","display_name":"Tao Zhang","orcid":"https://orcid.org/0009-0000-8302-1170","observed_orcids":["https://orcid.org/0009-0000-8302-1170"]},"institutions":[{"id":"https://openalex.org/I4405270328","display_name":"Ant Group (China)","ror":"https://ror.org/00g5w7n47","country_code":"CN","type":"company","lineage":["https://openalex.org/I4405270328"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tao Zhang","raw_affiliation_strings":["Ant Group, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0000-8302-1170","affiliations":[{"raw_affiliation_string":"Ant Group, Shanghai, China","institution_ids":["https://openalex.org/I4405270328"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.52106998,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4965","last_page":"4970"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9970999956130981,"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":0.9970999956130981,"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":0.000699999975040555,"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/T10203","display_name":"Recommender Systems and Techniques","score":0.000699999975040555,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/knowledge-distillation","display_name":"knowledge distillation","score":0.7839999794960022},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.7549999952316284},{"id":"https://openalex.org/keywords/chain-of-thought-reasoning","display_name":"chain-of-thought reasoning","score":0.6710000038146973},{"id":"https://openalex.org/keywords/group-relative-policy-optimization","display_name":"Group Relative Policy Optimization","score":0.8019999861717224},{"id":"https://openalex.org/keywords/token-level-supervision","display_name":"token-level supervision","score":0.7269999980926514},{"id":"https://openalex.org/keywords/search-engine-ranking","display_name":"search engine ranking","score":0.375}],"concepts":[{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.6646999716758728},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.638700008392334},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5835999846458435},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5548999905586243},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4767000079154968},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.45750001072883606},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.4487999975681305},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.35740000009536743},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3050000071525574},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.2809999883174896},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.28049999475479126},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.26080000400543213}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3805712.3808444","doi":"https://doi.org/10.1145/3805712.3808444","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805712.3808444","pdf_url":null,"source":{"id":"https://openalex.org/S4363608773","display_name":"International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3805712.3808444","doi":"https://doi.org/10.1145/3805712.3808444","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805712.3808444","pdf_url":null,"source":{"id":"https://openalex.org/S4363608773","display_name":"International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":3,"referenced_works":["https://openalex.org/W3021397474","https://openalex.org/W4252076394","https://openalex.org/W4400531953"],"related_works":["https://openalex.org/W4410636343","https://openalex.org/W4417194901","https://openalex.org/W7126163891","https://openalex.org/W4410636959","https://openalex.org/W7128634541","https://openalex.org/W4417011585","https://openalex.org/W4414990991","https://openalex.org/W4416400987","https://openalex.org/W7168349820","https://openalex.org/W4416401411"],"abstract_inverted_index":{"E-commerce":[0],"search":[1,224],"relevance":[2],"modeling":[3],"faces":[4],"a":[5,41,80,85,104,112,142,157,183,205,210],"critical":[6],"dilemma:":[7],"traditional":[8],"models":[9,94],"falter":[10],"with":[11,73,95,108],"complex":[12],"queries,":[13],"while":[14],"Large":[15],"Language":[16],"Models":[17],"(LLMs),":[18],"despite":[19],"their":[20],"superior":[21],"reasoning,":[22,86],"suffer":[23],"from":[24],"the":[25,52,129,180],"prohibitive":[26],"latency":[27],"of":[28,182,222],"auto-regressive":[29],"Chain-of-Thought":[30],"(CoT)":[31],"generation,":[32],"rendering":[33],"them":[34],"infeasible":[35],"for":[36,132],"production.":[37],"Knowledge":[38],"distillation":[39,162],"offers":[40],"promising":[42],"solution,":[43],"yet":[44],"current":[45],"methods":[46],"force":[47],"an":[48],"undesirable":[49],"trade-off:":[50],"sacrificing":[51],"very":[53],"interpretability":[54],"that":[55,83,163,174],"makes":[56],"LLMs":[57],"powerful,":[58],"or":[59],"relying":[60],"on":[61,226],"expensive,":[62],"unscalable":[63],"human-annotated":[64],"rationales.":[65],"To":[66],"address":[67],"this,":[68],"we":[69],"propose":[70],"STAR\u2014Staged":[71],"Training":[72],"Aligned":[74],"Reinforcement":[75],"Learning":[76],"and":[77,88,99,168,191,209],"Multi-Faceted":[78],"Distillation,":[79],"progressive":[81],"framework":[82],"follows":[84],"ranking,":[87],"transfer":[89],"pipeline":[90],"to":[91,156,220],"imbue":[92],"dense":[93],"both":[96,165],"high":[97],"performance":[98,181],"interpretability.":[100],"First,":[101],"STAR":[102,215],"aligns":[103],"teacher":[105],"LLM's":[106],"reasoning":[107,166],"task":[109],"objectives":[110],"using":[111],"novel":[113],"multi-granularity":[114],"reward":[115],"in":[116,213],"Group":[117],"Relative":[118],"Policy":[119],"Optimization":[120],"(GRPO),":[121],"leveraging":[122],"only":[123],"binary":[124],"labels.":[125],"Next,":[126],"it":[127,188],"refines":[128],"teacher's":[130],"ability":[131],"calibrated":[133],"scoring":[134],"via":[135,160],"token-level":[136],"supervision,":[137],"enabling":[138],"efficient":[139,190],"ranking":[140,169],"through":[141],"single":[143],"forward":[144],"pass":[145],"without":[146],"any":[147],"additional":[148],"layers.":[149],"Finally,":[150],"this":[151],"''white-box''":[152],"knowledge":[153],"is":[154,196],"transferred":[155],"compact":[158],"student":[159,177],"multi-faceted":[161],"preserves":[164],"logic":[167],"behavior.":[170],"Offline":[171],"experiments":[172],"demonstrate":[173],"our":[175],"0.6B":[176],"model":[178],"rivals":[179],"strong":[184],"8B":[185],"baseline,":[186],"making":[187],"highly":[189],"fully":[192,218],"deployable.":[193],"Real-world":[194],"effectiveness":[195],"validated":[197],"by":[198],"significant":[199],"online":[200],"A/B":[201],"test":[202],"gains,":[203],"including":[204],"+0.93%":[206],"GoodRate":[207],"lift":[208],"+1.04%":[211],"increase":[212],"GMV.":[214],"has":[216],"been":[217],"deployed":[219],"100%":[221],"main":[223],"traffic":[225],"1688.com.":[227]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-07-16T00:00:00"}