{"id":"https://openalex.org/W7130588722","doi":"https://doi.org/10.48550/arxiv.2602.15856","title":"Rethinking Soft Compression in Retrieval-Augmented Generation: A Query-Conditioned Selector Perspective","display_name":"Rethinking Soft Compression in Retrieval-Augmented Generation: A Query-Conditioned Selector Perspective","publication_year":2026,"publication_date":"2026-01-25","ids":{"openalex":"https://openalex.org/W7130588722","doi":"https://doi.org/10.48550/arxiv.2602.15856"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2602.15856","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.15856","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2602.15856","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5126383266","display_name":"Yunhao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Liu, Yunhao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041693857","display_name":"Zian Jia","orcid":"https://orcid.org/0000-0002-0181-9961"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jia, Zian","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124120364","display_name":"Xinyu Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Xinyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126409184","display_name":"Kanjun Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Kanjun","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5126401224","display_name":"Yun Xiong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiong, Yun","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5126383266"],"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":0.26429998874664307,"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":0.26429998874664307,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.1559000015258789,"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":0.12300000339746475,"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/encoding","display_name":"Encoding (memory)","score":0.6352999806404114},{"id":"https://openalex.org/keywords/relevance","display_name":"Relevance (law)","score":0.579200029373169},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5713000297546387},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5687000155448914},{"id":"https://openalex.org/keywords/compression","display_name":"Compression (physics)","score":0.5407000184059143},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.5128999948501587},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.49790000915527344}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7648000121116638},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.6352999806404114},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.579200029373169},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5713000297546387},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5687000155448914},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.5407000184059143},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.5128999948501587},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.49790000915527344},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4553999900817871},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.43389999866485596},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4318000078201294},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.33160001039505005},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3192000091075897},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.3052000105381012},{"id":"https://openalex.org/C180198813","wikidata":"https://www.wikidata.org/wiki/Q121182","display_name":"Information system","level":2,"score":0.29490000009536743},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.28450000286102295},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.2728999853134155},{"id":"https://openalex.org/C2781181686","wikidata":"https://www.wikidata.org/wiki/Q4226068","display_name":"Coherence (philosophical gambling strategy)","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.25859999656677246},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2517000138759613}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2602.15856","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.15856","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2602.15856","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.15856","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":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.7503854036331177}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Retrieval-Augmented":[0],"Generation":[1],"(RAG)":[2],"effectively":[3],"grounds":[4],"Large":[5],"Language":[6],"Models":[7],"(LLMs)":[8],"with":[9,96,143,152],"external":[10],"knowledge":[11],"and":[12,28,87,102,108,140,147,166,178],"is":[13,22,106,138,141],"widely":[14],"applied":[15],"to":[16,38,55,65,73,172],"Web-related":[17],"tasks.":[18],"However,":[19],"its":[20],"scalability":[21],"hindered":[23],"by":[24,41,114,180],"excessive":[25],"context":[26,35],"length":[27],"redundant":[29],"retrievals.":[30],"Recent":[31],"research":[32],"on":[33,58,84],"soft":[34,122,163],"compression":[36,123,164],"aims":[37],"address":[39],"this":[40,78,85],"encoding":[42],"long":[43],"documents":[44],"into":[45],"compact":[46],"embeddings,":[47],"yet":[48],"they":[49],"often":[50],"underperform":[51],"non-compressed":[52],"RAG":[53,126],"due":[54],"their":[56],"reliance":[57],"auto-encoder-like":[59],"full-compression":[60,94,105],"that":[61,127,158],"forces":[62],"the":[63,74,97,129],"encoder":[64],"compress":[66],"all":[67],"document":[68],"information":[69,111,134],"regardless":[70],"of":[71],"relevance":[72],"input":[75],"query.":[76],"In":[77],"work,":[79],"we":[80,117],"conduct":[81],"an":[82],"analysis":[83],"paradigm":[86],"reveal":[88],"two":[89],"fundamental":[90],"limitations:":[91],"(I)":[92],"Infeasibility,":[93],"conflicts":[95],"LLM's":[98],"downstream":[99],"generation":[100],"behavior;":[101],"(II)":[103],"Non-necessity:":[104],"unnecessary":[107],"dilutes":[109],"task-relevant":[110],"density.":[112],"Motivated":[113],"these":[115],"insights,":[116],"introduce":[118],"SeleCom,":[119],"a":[120,144],"selector-based":[121],"framework":[124],"for":[125],"redefines":[128],"encoder's":[130],"role":[131],"as":[132],"query-conditioned":[133],"selector.":[135],"The":[136],"selector":[137],"decoder-only":[139],"trained":[142],"massive,":[145],"diverse":[146],"difficulty-graded":[148],"synthetic":[149],"QA":[150],"dataset":[151],"curriculum":[153],"learning.":[154],"Extensive":[155],"experiments":[156],"show":[157],"SeleCom":[159],"significantly":[160],"outperforms":[161],"existing":[162],"approaches":[165],"achieves":[167],"competitive":[168],"or":[169],"superior":[170],"performance":[171],"non-compression":[173],"baselines,":[174],"while":[175],"reducing":[176],"computation":[177],"latency":[179],"33.8%~84.6%.":[181]},"counts_by_year":[],"updated_date":"2026-02-20T06:18:38.638704","created_date":"2026-02-20T00:00:00"}