{"id":"https://openalex.org/W7164942249","doi":"https://doi.org/10.48550/arxiv.2606.16409","title":"PathRouter: Aligning Rewards with Retrieval Quality in Agentic Graph Retrieval-Augmented Generation","display_name":"PathRouter: Aligning Rewards with Retrieval Quality in Agentic Graph Retrieval-Augmented Generation","publication_year":2026,"publication_date":"2026-06-15","ids":{"openalex":"https://openalex.org/W7164942249","doi":"https://doi.org/10.48550/arxiv.2606.16409"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.16409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.16409","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138721345","display_name":"Bo Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122964295","display_name":"H Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Heyan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030499518","display_name":"Yaolin Li","orcid":"https://orcid.org/0009-0006-3109-8137"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yaolin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138729139","display_name":"Wei Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138743933","display_name":"Yuan Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138738509","display_name":"Wenbo Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Wenbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138735453","display_name":"Mingze Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Mingze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138738947","display_name":"Ge Shi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shi, Ge","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138749007","display_name":"Chong Feng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Feng, Chong","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.4596000015735626,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.4596000015735626,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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":0.34790000319480896,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.12800000607967377,"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/correctness","display_name":"Correctness","score":0.7190999984741211},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.6288999915122986},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.48030000925064087},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.47690001130104065},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.45239999890327454}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7603999972343445},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.7190999984741211},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6288999915122986},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4918999969959259},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.48030000925064087},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.47690001130104065},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4618000090122223},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.45239999890327454},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.3400000035762787},{"id":"https://openalex.org/C190839683","wikidata":"https://www.wikidata.org/wiki/Q2448197","display_name":"Train","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.2768000066280365},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.16409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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":"doi:10.48550/arxiv.2606.16409","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16409","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"sustainable_development_goals":[{"score":0.42757293581962585,"id":"https://metadata.un.org/sdg/16","display_name":"Peace, Justice and strong institutions"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Agentic":[0],"GraphRAG":[1],"trains":[2],"language-model":[3],"agents":[4],"to":[5,61,124,162],"iteratively":[6],"retrieve":[7],"and":[8,16,85,118,145,156],"reason":[9],"over":[10],"graph-structured":[11],"evidence,":[12],"enabling":[13],"more":[14],"accurate":[15],"context-aware":[17],"decision-making":[18],"by":[19],"efficiently":[20],"navigating":[21],"complex":[22],"information":[23],"networks.":[24],"However,":[25],"outcome-only":[26],"reinforcement":[27,100],"learning":[28],"suffers":[29],"from":[30,39],"\\textit{\\textbf{answer-path":[31],"reward":[32],"aliasing}},":[33],"where":[34],"correct":[35],"answers":[36],"may":[37],"come":[38],"shortcuts":[40],"rather":[41],"than":[42],"useful":[43],"evidence":[44],"paths.":[45],"It":[46],"also":[47],"exhibits":[48],"\\textit{\\textbf{search-update":[49],"ambiguity}},":[50],"as":[51],"scalar":[52],"trajectory-level":[53],"feedback":[54],"does":[55],"not":[56],"indicate":[57],"which":[58],"retrieval":[59],"actions":[60],"adjust.":[62],"To":[63],"mitigate":[64],"these":[65],"shortcomings,":[66],"we":[67],"present":[68],"PathRouter,":[69],"a":[70,108,163],"path-aware":[71],"training":[72],"framework":[73],"for":[74],"agentic":[75],"GraphRAG.":[76],"PathRouter":[77,140],"jointly":[78],"evaluates":[79],"each":[80],"trajectory":[81,90],"along":[82],"answer":[83,122,143],"correctness":[84],"evidence-path":[86,146],"overlap,":[87,147],"yielding":[88],"four":[89],"categories":[91],"with":[92],"differentiated":[93],"GRPO":[94],"advantage":[95],"scaling":[96],"that":[97,139],"suppresses":[98],"shortcut":[99],"while":[101],"preserving":[102],"evidence-seeking":[103],"behavior.":[104],"For":[105],"evidence-poor":[106],"trajectories,":[107],"frozen":[109],"gold-evidence":[110],"teacher":[111],"provides":[112],"token-level":[113],"KL":[114],"guidance":[115],"on":[116,130,154,158],"reasoning":[117],"search-query":[119],"tokens,":[120],"excluding":[121],"tokens":[123],"avoid":[125],"direct":[126],"response":[127],"imitation.":[128],"Experiments":[129],"six":[131],"QA":[132],"benchmarks":[133],"across":[134],"three":[135],"model":[136],"sizes":[137],"show":[138],"consistently":[141],"improves":[142],"F1":[144,150],"achieving":[148],"average":[149],"gains":[151],"of":[152],"3.1":[153],"3B":[155],"4.9":[157],"7B":[159],"models":[160],"compared":[161],"strong":[164],"baseline.":[165]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-17T00:00:00"}