{"id":"https://openalex.org/W7166804173","doi":"https://doi.org/10.18653/v1/2026.semeval-1.416","title":"RAGthoven at SemEval-2026 Task 1: A Multi-Stage Pipeline Walks Into a Benchmark and Barely Clears the Bar","display_name":"RAGthoven at SemEval-2026 Task 1: A Multi-Stage Pipeline Walks Into a Benchmark and Barely Clears the Bar","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166804173","doi":"https://doi.org/10.18653/v1/2026.semeval-1.416"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.semeval-1.416","doi":"https://doi.org/10.18653/v1/2026.semeval-1.416","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.semeval-1.416","pdf_url":"https://aclanthology.org/2026.semeval-1.416.pdf","source":{"id":"https://openalex.org/S4363608670","display_name":"International Workshop on Semantic Evaluation (SemEval)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 20th International Workshop on Semantic Evaluation (2026)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2026.semeval-1.416.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5124522043","display_name":"Marek Suppa","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Marek Suppa","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5105957331","display_name":"Vikt\u00f3ria Ondrejov\u00e1","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vikt\u00f3ria Ondrejov\u00e1","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139824890","display_name":"Lucia Ganajov\u00e1","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lucia Ganajov\u00e1","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044304429","display_name":"Gregor Karetka","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gregor Karetka","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5113067493","display_name":"Daniel Skala","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Daniel Skala","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":0.44083095,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3343","last_page":"3356"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11795","display_name":"Humor Studies and Applications","score":0.6812000274658203,"subfield":{"id":"https://openalex.org/subfields/3207","display_name":"Social Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T11795","display_name":"Humor Studies and Applications","score":0.6812000274658203,"subfield":{"id":"https://openalex.org/subfields/3207","display_name":"Social Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.22830000519752502,"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":0.03060000017285347,"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/multilingual-text-generation","display_name":"multilingual text generation","score":0.703000009059906},{"id":"https://openalex.org/keywords/retrieval-augmented-generation","display_name":"retrieval-augmented generation","score":0.9380000233650208},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.625},{"id":"https://openalex.org/keywords/benign-violation-theory","display_name":"benign violation theory","score":0.8730000257492065},{"id":"https://openalex.org/keywords/prompt-engineering","display_name":"prompt engineering","score":0.847000002861023},{"id":"https://openalex.org/keywords/agentic-llms","display_name":"agentic LLMs","score":0.4869999885559082},{"id":"https://openalex.org/keywords/tool-calling","display_name":"tool calling","score":0.718999981880188}],"concepts":[{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.7027000188827515},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6660000085830688},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6434000134468079},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6291000247001648},{"id":"https://openalex.org/C188721877","wikidata":"https://www.wikidata.org/wiki/Q103510","display_name":"Bar (unit)","level":2,"score":0.5595999956130981},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.305400013923645},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.29089999198913574},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2892000079154968},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.28529998660087585},{"id":"https://openalex.org/C199639397","wikidata":"https://www.wikidata.org/wiki/Q1788588","display_name":"Engineering drawing","level":1,"score":0.2712000012397766},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.26330000162124634}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.semeval-1.416","doi":"https://doi.org/10.18653/v1/2026.semeval-1.416","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.semeval-1.416","pdf_url":"https://aclanthology.org/2026.semeval-1.416.pdf","source":{"id":"https://openalex.org/S4363608670","display_name":"International Workshop on Semantic Evaluation (SemEval)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 20th International Workshop on Semantic Evaluation (2026)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.semeval-1.416","doi":"https://doi.org/10.18653/v1/2026.semeval-1.416","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.semeval-1.416","pdf_url":"https://aclanthology.org/2026.semeval-1.416.pdf","source":{"id":"https://openalex.org/S4363608670","display_name":"International Workshop on Semantic Evaluation (SemEval)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 20th International Workshop on Semantic Evaluation (2026)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166804173.pdf","grobid_xml":"https://content.openalex.org/works/W7166804173.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7166883141","https://openalex.org/W7166874256","https://openalex.org/W7166844705","https://openalex.org/W7166810646","https://openalex.org/W7166821275","https://openalex.org/W7166801503","https://openalex.org/W7166858953","https://openalex.org/W7166870295","https://openalex.org/W7166859713","https://openalex.org/W7166806194"],"abstract_inverted_index":{"We":[0],"present":[1],"RAGTHOVEN,":[2],"our":[3,59],"system":[4,151],"for":[5],"SemEval-2026":[6],"Task":[7],"1":[8,128,161],"(MuWaHaHa),":[9],"Subtask":[10],"A":[11],"(multilingual":[12],"constrained":[13],"humor":[14,40],"generation":[15,24,68,76],"in":[16,38,129,137,153,156,165,174,198],"English,":[17],"Spanish,":[18,157],"and":[19,50,176],"Chinese).RAGTHOVEN":[20],"decomposes":[21],"creative":[22],"text":[23],"into":[25],"a":[26,71,98,102,117,193],"multi-stage":[27,185],"large":[28],"language":[29],"model":[30,99,196],"(LLM)":[31],"pipeline":[32,91],"(Planner,":[33],"Writer,":[34],"Reflector,":[35],"Judge)":[36],"grounded":[37],"computational":[39],"theories":[41],"(Benign":[42],"Violation":[43],"Theory,":[44],"Script-based":[45],"Semantic":[46],"Theory":[47],"of":[48],"Humor)":[49],"iteratively":[51],"refined":[52],"through":[53],"prompt":[54,186],"engineering":[55,187],"across":[56],"ten":[57],"experiments.In":[58],"final":[60],"configuration,":[61],"we":[62],"augment":[63],"the":[64,88,122,134,144,150,163,178,199],"Planner":[65],"with":[66,77,101,133,162],"retrieval-augmented":[67],"(RAG)":[69],"from":[70],"curated":[72],"joke":[73,79],"corpus,":[74],"seeding":[75],"diverse":[78],"mechanisms.We":[80],"additionally":[81],"explore":[82],"an":[83],"agentic":[84],"variant":[85],"that":[86,183],"exposes":[87],"same":[89],"four":[90],"stages":[92],"as":[93],"toolcalling":[94],"agents":[95],"orchestrated":[96],"by":[97],"loop":[100],"CONSTRAINTAUDIT":[103],"checker;":[104],"while":[105,149],"it":[106,158],"achieves":[107,126],"full":[108],"constraint":[109],"compliance,":[110],"human":[111],"pairwise":[112],"evaluation":[113],"did":[114],"not":[115],"reveal":[116],"significant":[118],"quality":[119],"advantage":[120],"over":[121],"simpler":[123],"non-agentic":[124],"baseline.RAGTHOVEN":[125],"Rank":[127,160],"all":[130,166],"three":[131,167],"languages,":[132],"strongest":[135],"result":[136],"Spanish":[138],"(Elo":[139],"1182,":[140],"42":[141],"points":[142],"above":[143],"Gemini":[145],"2.5":[146],"Flash":[147],"baseline).However,":[148],"leads":[152],"raw":[154],"Elo":[155],"shares":[159],"baseline":[164],"languages":[168],"due":[169],"to":[170],"overlapping":[171],"confidence":[172],"intervals;":[173],"English":[175],"Chinese":[177],"gap":[179],"narrows":[180],"further,":[181],"suggesting":[182],"elaborate":[184],"may":[188],"offer":[189],"diminishing":[190],"returns":[191],"once":[192],"strong":[194],"frontier":[195],"is":[197],"loop.":[200]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-07-02T00:00:00"}