{"id":"https://openalex.org/W7154513540","doi":"https://doi.org/10.48550/arxiv.2604.12231","title":"Thought-Retriever: Don't Just Retrieve Raw Data, Retrieve Thoughts for Memory-Augmented Agentic Systems","display_name":"Thought-Retriever: Don't Just Retrieve Raw Data, Retrieve Thoughts for Memory-Augmented Agentic Systems","publication_year":2026,"publication_date":"2026-04-14","ids":{"openalex":"https://openalex.org/W7154513540","doi":"https://doi.org/10.48550/arxiv.2604.12231"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.12231","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12231","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":"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.2604.12231","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100724815","display_name":"Tao Feng","orcid":"https://orcid.org/0000-0003-2247-0751"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Feng, Tao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133691938","display_name":"Pengrui Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Pengrui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133675082","display_name":"Guanyu Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Guanyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133721550","display_name":"Ge Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Ge","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133639379","display_name":"Jiaxuan You","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"You, Jiaxuan","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":0.17399999499320984,"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":0.17399999499320984,"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/T14347","display_name":"Big Data and Digital Economy","score":0.1615000069141388,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.12330000102519989,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.8791999816894531},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5622000098228455},{"id":"https://openalex.org/keywords/raw-data","display_name":"Raw data","score":0.5569999814033508},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5281999707221985},{"id":"https://openalex.org/keywords/knowledge-base","display_name":"Knowledge base","score":0.39480000734329224}],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.8791999816894531},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7671999931335449},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5622000098228455},{"id":"https://openalex.org/C132964779","wikidata":"https://www.wikidata.org/wiki/Q2110223","display_name":"Raw data","level":2,"score":0.5569999814033508},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5281999707221985},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.4187000095844269},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.39480000734329224},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3785000145435333},{"id":"https://openalex.org/C2776145971","wikidata":"https://www.wikidata.org/wiki/Q30673951","display_name":"Labeled data","level":2,"score":0.37290000915527344},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3009999990463257},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.28189998865127563},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C136764020","wikidata":"https://www.wikidata.org/wiki/Q466","display_name":"World Wide Web","level":1,"score":0.27230000495910645},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.25600001215934753},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.25540000200271606}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.12231","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12231","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":"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.2604.12231","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12231","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":"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":[{"id":"https://metadata.un.org/sdg/9","display_name":"Industry, innovation and infrastructure","score":0.6366593241691589}],"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],"transformed":[5],"AI":[6],"research":[7],"thanks":[8],"to":[9,21,38,111,178,183,249,253],"their":[10],"powerful":[11],"internal":[12],"capabilities":[13],"and":[14,129,137,195,218],"knowledge.":[15],"However,":[16],"existing":[17],"LLMs":[18,35,85],"still":[19,44],"fail":[20],"effectively":[22,147],"incorporate":[23],"the":[24,31,40,48,63,98,139],"massive":[25],"external":[26,64,92],"knowledge":[27,65],"when":[28,121,142],"interacting":[29],"with":[30,151],"world.":[32],"Although":[33],"retrieval-augmented":[34],"are":[36,43],"proposed":[37],"mitigate":[39],"issue,":[41],"they":[42,54],"fundamentally":[45],"constrained":[46,96],"by":[47,97],"context":[49,99,182],"length":[50,100],"of":[51,70,72,103,211],"LLMs,":[52],"as":[53],"can":[55,236],"only":[56],"retrieve":[57],"top-K":[58],"raw":[59],"data":[60,73,105],"chunks":[61],"from":[62],"base":[66],"which":[67,174],"often":[68],"consists":[69],"millions":[71],"chunks.":[74,106],"Here":[75],"we":[76,166,228],"propose":[77],"Thought-Retriever,":[78],"a":[79,152,170],"novel":[80,171],"model-agnostic":[81],"algorithm":[82],"that":[83,156,201],"helps":[84],"generate":[86],"output":[87],"conditioned":[88],"on":[89,187,193],"arbitrarily":[90],"long":[91],"data,":[93],"without":[94],"being":[95],"or":[101],"number":[102],"retrieved":[104],"Our":[107],"key":[108],"insight":[109],"is":[110],"let":[112],"an":[113,176,208],"LLM":[114,177,239],"fully":[115],"leverage":[116,180,250],"its":[117],"intermediate":[118],"responses":[119],"generated":[120],"solving":[122,242],"past":[123],"user":[124,244,257],"queries":[125,185],"(thoughts),":[126],"filtering":[127],"meaningless":[128],"redundant":[130],"thoughts,":[131],"organizing":[132],"them":[133],"in":[134,215,220],"thought":[135],"memory,":[136],"retrieving":[138],"relevant":[140],"thoughts":[141,252],"addressing":[143],"new":[144],"queries.":[145,258],"This":[146],"equips":[148],"LLM-based":[149],"agents":[150],"self-evolving":[153],"long-term":[154],"memory":[155],"grows":[157],"more":[158,243,255],"capable":[159],"through":[160],"continuous":[161],"interaction.":[162],"Besides":[163],"algorithmic":[164],"innovation,":[165],"further":[167,229],"meticulously":[168],"prepare":[169],"benchmark,":[172],"AcademicEval,":[173],"requires":[175],"faithfully":[179],"ultra-long":[181],"answer":[184,254],"based":[186],"real-world":[188],"academic":[189],"papers.":[190],"Extensive":[191],"experiments":[192],"AcademicEval":[194],"two":[196,231],"other":[197],"public":[198],"datasets":[199],"validate":[200],"Thought-Retriever":[202,235,247],"remarkably":[203],"outperforms":[204],"state-of-the-art":[205],"baselines,":[206],"achieving":[207],"average":[209],"increase":[210],"at":[212],"least":[213],"7.6%":[214],"F1":[216],"score":[217],"16%":[219],"win":[221],"rate":[222],"across":[223],"various":[224],"tasks.":[225],"More":[226],"importantly,":[227],"demonstrate":[230],"exciting":[232],"findings:":[233],"(1)":[234],"indeed":[237],"help":[238],"self-evolve":[240],"after":[241],"queries;":[245],"(2)":[246],"learns":[248],"deeper":[251],"abstract":[256]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-04-16T00:00:00"}