{"id":"https://openalex.org/W7166809132","doi":"https://doi.org/10.18653/v1/2026.acl-long.1462","title":"Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization","display_name":"Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166809132","doi":"https://doi.org/10.18653/v1/2026.acl-long.1462"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1462","pdf_url":"https://aclanthology.org/2026.acl-long.1462.pdf","source":null,"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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.1462.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114723683","display_name":"Junzhe Wang","orcid":"https://orcid.org/0009-0006-8449-5419"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Junzhe Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139764928","display_name":"Zhiheng Xi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiheng Xi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139789068","display_name":"Yajie Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yajie Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139849931","display_name":"Hao Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao Luo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139758494","display_name":"Shihan Dou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shihan Dou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139809977","display_name":"Tao Gui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tao Gui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139758763","display_name":"Qi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qi Zhang","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.79904777,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"31704","last_page":"31718"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.14650000631809235,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.14650000631809235,"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/T12288","display_name":"Optimization and Search Problems","score":0.13359999656677246,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.07190000265836716,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/minification","display_name":"Minification","score":0.3292999863624573},{"id":"https://openalex.org/keywords/group","display_name":"Group (periodic table)","score":0.3077000081539154},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.25940001010894775},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.25679999589920044}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.498199999332428},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3544999957084656},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.3292999863624573},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.31139999628067017},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.30869999527931213},{"id":"https://openalex.org/C2781311116","wikidata":"https://www.wikidata.org/wiki/Q83306","display_name":"Group (periodic table)","level":2,"score":0.3077000081539154},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2651999890804291},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.25679999589920044},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.24609999358654022}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1462","pdf_url":"https://aclanthology.org/2026.acl-long.1462.pdf","source":null,"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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.1462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1462","pdf_url":"https://aclanthology.org/2026.acl-long.1462.pdf","source":null,"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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1187148928","display_name":null,"funder_award_id":"62376061","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G5542150374","display_name":null,"funder_award_id":"62476061","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166809132.pdf","grobid_xml":"https://content.openalex.org/works/W7166809132.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Search":[0],"agents":[1],"extend":[2],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"beyond":[7],"static":[8],"parametric":[9],"knowledge":[10],"by":[11,130],"enabling":[12,112],"access":[13],"to":[14,53,86,105,138],"up-to-date":[15],"and":[16,91,134],"longtail":[17],"information":[18],"unavailable":[19],"during":[20],"pretraining.While":[21],"reinforcement":[22],"learning":[23],"has":[24],"been":[25],"widely":[26],"adopted":[27],"for":[28],"training":[29],"such":[30],"agents,":[31],"existing":[32],"approaches":[33],"face":[34],"key":[35],"limitations:":[36],"process":[37,69,79],"supervision":[38,47,70],"often":[39],"suffers":[40],"from":[41],"unstable":[42],"value":[43],"estimation,":[44],"whereas":[45],"outcome":[46],"struggles":[48],"with":[49],"credit":[50,114],"assignment":[51,115],"due":[52],"sparse,":[54],"trajectory-level":[55],"rewards.To":[56],"bridge":[57],"this":[58],"gap,":[59],"we":[60],"propose":[61],"Contribution-Weighted":[62],"GRPO":[63,129],"(CW-GRPO),":[64],"a":[65],"framework":[66],"that":[67,125,145],"integrates":[68],"into":[71,157],"group":[72],"relative":[73],"policy":[74],"optimization.Instead":[75],"of":[76],"directly":[77],"optimizing":[78],"rewards,":[80],"CW-GRPO":[81,126],"employs":[82],"an":[83],"LLM":[84],"judge":[85],"assess":[87],"the":[88,110],"retrieval":[89],"utility":[90],"reasoning":[92],"correctness":[93],"at":[94,164],"each":[95],"search":[96,141,158],"round,":[97],"producing":[98],"per-round":[99],"contribution":[100],"weights.These":[101],"weights":[102],"are":[103],"used":[104],"rescale":[106],"outcome-based":[107],"advantages":[108],"along":[109],"trajectory,":[111],"finegrained":[113],"without":[116],"sacrificing":[117],"optimization":[118],"stability.Experiments":[119],"on":[120,132,136],"multiple":[121],"knowledge-intensive":[122],"benchmarks":[123],"show":[124],"outperforms":[127],"standard":[128],"5.0%":[131],"Qwen3-8B":[133],"6.3%":[135],"Qwen3-1.7B,leading":[137],"more":[139],"effective":[140],"behaviors.Additional":[142],"analysis":[143],"reveals":[144],"successful":[146],"trajectories":[147],"exhibit":[148],"concentrated":[149],"contributions":[150],"in":[151],"specific":[152],"rounds,":[153],"providing":[154],"empirical":[155],"insight":[156],"agent":[159],"tasks.Our":[160],"code":[161],"is":[162],"available":[163],"https://github.com/zsxmwjz/CW-GRPO.":[165]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}