{"id":"https://openalex.org/W4416034645","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.504","title":"Efficient Integration of External Knowledge to LLM-based World Models via Retrieval-Augmented Generation and Reinforcement Learning","display_name":"Efficient Integration of External Knowledge to LLM-based World Models via Retrieval-Augmented Generation and Reinforcement Learning","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034645","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.504"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.504","pdf_url":"https://aclanthology.org/2025.findings-emnlp.504.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.504.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113820665","display_name":"Chang Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chang Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049783685","display_name":"Xinrun Wang","orcid":"https://orcid.org/0000-0003-3369-219X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinrun Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058417704","display_name":"Qinggang Zhang","orcid":"https://orcid.org/0000-0002-0247-4942"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qinggang Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048899273","display_name":"Qi Jiang","orcid":"https://orcid.org/0009-0007-0828-1504"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qi Jiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Xiao Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao Huang","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":0.29935113,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"9484","last_page":"9501"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.3716000020503998,"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":0.3716000020503998,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.07989999651908875,"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":0.05770000070333481,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.5008999705314636},{"id":"https://openalex.org/keywords/control","display_name":"Control (management)","score":0.3450999855995178},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.31139999628067017},{"id":"https://openalex.org/keywords/knowledge-based-systems","display_name":"Knowledge-based systems","score":0.27630001306533813},{"id":"https://openalex.org/keywords/field","display_name":"Field (mathematics)","score":0.2712000012397766},{"id":"https://openalex.org/keywords/production","display_name":"Production (economics)","score":0.26109999418258667}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5931000113487244},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5008999705314636},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4690999984741211},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3450999855995178},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.31139999628067017},{"id":"https://openalex.org/C115925183","wikidata":"https://www.wikidata.org/wiki/Q1412694","display_name":"Knowledge-based systems","level":2,"score":0.27630001306533813},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2712000012397766},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2612999975681305},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2578999996185303},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2574999928474426},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2533999979496002},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.25049999356269836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.504","pdf_url":"https://aclanthology.org/2025.findings-emnlp.504.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.504","pdf_url":"https://aclanthology.org/2025.findings-emnlp.504.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320322724","display_name":"Ministry of Education, India","ror":"https://ror.org/048xjjh50"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034645.pdf","grobid_xml":"https://content.openalex.org/works/W4416034645.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"World":[0,65],"models":[1,47,161],"achieve":[2],"remarkable":[3],"success":[4],"in":[5,11],"predicting":[6],"future":[7],"states":[8],"and":[9,14,94,153,162,170,182,194],"planning":[10],"complex":[12],"environments,":[13,149],"Large":[15],"Language":[16],"Models":[17,66],"(LLMs)":[18],"serve":[19],"as":[20,103],"promising":[21],"foundations":[22],"to":[23,38,43,75,99,107],"build":[24],"general":[25],"world":[26,46,82,110],"models.However,":[27],"their":[28],"performances":[29],"are":[30,54,86],"usually":[31],"constrained":[32],"by":[33],"the":[34,104,109,129,167,171],"limited":[35],"external":[36,78,187],"knowledge":[37,58,79,188],"specific":[39],"environments.Existing":[40],"research":[41],"attempts":[42],"enhance":[44],"LLM-based":[45,81],"through":[48],"prompting":[49],"or":[50,59],"fine-tuning":[51],"approaches,":[52],"which":[53],"either":[55],"requiring":[56],"human":[57],"computationally":[60],"extensive.Therefore,":[61],"we":[62],"introduce":[63,90],"Retrieval-Augmented":[64],"(RAWM),":[67],"a":[68,91,116,124],"novel":[69],"framework":[70],"that":[71,122,156],"leverages":[72],"retrievalaugmented":[73],"generation":[74,169],"efficiently":[76],"integrate":[77],"into":[80],"models.Our":[83],"main":[84],"contributions":[85],"threefold:":[87],"(i)":[88],"We":[89,114,142],"memory":[92],"system":[93],"design":[95],"an":[96],"embedding":[97,131],"model":[98,132],"retrieve":[100],"relevant":[101,179],"experiences":[102,181],"in-context":[105],"examples":[106],"improve":[108],"model's":[111],"predictive":[112],"accuracy.(ii)":[113],"develop":[115],"reinforcement":[117],"learning":[118],"(RL)":[119],"training":[120,174],"pipeline":[121],"fine-tunes":[123],"small":[125],"MLP":[126],"head":[127],"on":[128],"pre-trained":[130],"using":[133],"Proximal":[134],"Policy":[135],"Optimization":[136],"(PPO),":[137],"further":[138],"enhancing":[139],"prediction":[140],"performance.(iii)":[141],"conduct":[143],"extensive":[144],"experiments":[145],"across":[146],"three":[147],"diverse":[148],"i.e.,":[150],"Game24,":[151],"BlocksWorld,":[152],"BabyAI,":[154],"demonstrating":[155],"RAWM":[157,176],"consistently":[158],"outperforms":[159],"baseline":[160],"exhibits":[163],"strong":[164],"generalizability.By":[165],"leveraging":[166],"retrieval-augmented":[168],"efficient":[172],"RL":[173],"pipeline,":[175],"dynamically":[177],"utilizes":[178],"historical":[180],"equips":[183],"LLMs":[184],"with":[185],"environment-specific":[186],"without":[189],"retraining,":[190],"enabling":[191],"more":[192],"accurate":[193],"generalizable":[195],"predictions.":[196]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-11-08T00:00:00"}