{"id":"https://openalex.org/W4401042124","doi":"https://doi.org/10.18653/v1/2024.semeval-1.191","title":"HaRMoNEE at SemEval-2024 Task 6: Tuning-based Approaches to Hallucination Recognition","display_name":"HaRMoNEE at SemEval-2024 Task 6: Tuning-based Approaches to Hallucination Recognition","publication_year":2024,"publication_date":"2024-01-01","ids":{"openalex":"https://openalex.org/W4401042124","doi":"https://doi.org/10.18653/v1/2024.semeval-1.191"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2024.semeval-1.191","doi":"https://doi.org/10.18653/v1/2024.semeval-1.191","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.semeval-1.191","pdf_url":"https://aclanthology.org/2024.semeval-1.191.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 18th International Workshop on Semantic Evaluation (SemEval-2024)","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/2024.semeval-1.191.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5106203187","display_name":"Timothy Obiso","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I6902469","display_name":"Brandeis University","ror":"https://ror.org/05abbep66","country_code":"US","type":"education","lineage":["https://openalex.org/I6902469"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Timothy Obiso","raw_affiliation_strings":["Department of Computer Science Brandeis University Waltham , Massachusetts"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science Brandeis University Waltham , Massachusetts","institution_ids":["https://openalex.org/I6902469"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038194411","display_name":"Jingxuan Tu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I6902469","display_name":"Brandeis University","ror":"https://ror.org/05abbep66","country_code":"US","type":"education","lineage":["https://openalex.org/I6902469"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jingxuan Tu","raw_affiliation_strings":["Department of Computer Science Brandeis University Waltham , Massachusetts"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science Brandeis University Waltham , Massachusetts","institution_ids":["https://openalex.org/I6902469"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5012141433","display_name":"James D. Pustejovsky","orcid":"https://orcid.org/0000-0003-2233-9761","observed_orcids":["https://orcid.org/0000-0003-2233-9761"]},"institutions":[{"id":"https://openalex.org/I6902469","display_name":"Brandeis University","ror":"https://ror.org/05abbep66","country_code":"US","type":"education","lineage":["https://openalex.org/I6902469"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"James Pustejovsky","raw_affiliation_strings":["Department of Computer Science Brandeis University Waltham , Massachusetts"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science Brandeis University Waltham , Massachusetts","institution_ids":["https://openalex.org/I6902469"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I6902469"],"apc_list":null,"apc_paid":null,"fwci":0.2417,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.42790514,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":92,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"1322","last_page":"1331"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.8901000022888184,"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":0.8901000022888184,"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":0.06989999860525131,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.007600000128149986,"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/hallucination-detection","display_name":"hallucination detection","score":0.9850000143051147},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8289999961853027},{"id":"https://openalex.org/keywords/overgeneration","display_name":"overgeneration","score":0.5090000033378601},{"id":"https://openalex.org/keywords/fine-tuning","display_name":"fine-tuning","score":0.8050000071525574},{"id":"https://openalex.org/keywords/prompt-tuning","display_name":"prompt tuning","score":0.9959999918937683},{"id":"https://openalex.org/keywords/transformer","display_name":"transformer models","score":0.6480000019073486}],"concepts":[{"id":"https://openalex.org/C44572571","wikidata":"https://www.wikidata.org/wiki/Q7448970","display_name":"SemEval","level":3,"score":0.928866446018219},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7500019073486328},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.7439788579940796},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4914788007736206},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4577452540397644},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3254537582397461},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.05919906497001648},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2024.semeval-1.191","doi":"https://doi.org/10.18653/v1/2024.semeval-1.191","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.semeval-1.191","pdf_url":"https://aclanthology.org/2024.semeval-1.191.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 18th International Workshop on Semantic Evaluation (SemEval-2024)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2024.semeval-1.191","doi":"https://doi.org/10.18653/v1/2024.semeval-1.191","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.semeval-1.191","pdf_url":"https://aclanthology.org/2024.semeval-1.191.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 18th International Workshop on Semantic Evaluation (SemEval-2024)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G8418166701","display_name":"EAGER: Integrating Dense Paraphrased-Enriched Representations with Large Language Models","funder_award_id":"2326985","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4401042124.pdf","grobid_xml":"https://content.openalex.org/works/W4401042124.grobid-xml"},"referenced_works_count":1,"referenced_works":["https://openalex.org/W4389520749"],"related_works":["https://openalex.org/W4401042144","https://openalex.org/W4401042373","https://openalex.org/W4401042493","https://openalex.org/W4401043862","https://openalex.org/W7218038275","https://openalex.org/W4401043384","https://openalex.org/W4401043554","https://openalex.org/W4401042171","https://openalex.org/W4394780427","https://openalex.org/W4416542655"],"abstract_inverted_index":{"This":[0],"paper":[1],"presents":[2],"the":[3,70,78,84,88,92,95],"Hallucination":[4],"Recognition":[5],"Model":[6,46],"for":[7,18,77,94],"New":[8],"Experiment":[9],"Evaluation":[10],"(HaRMoNEE)":[11],"team's":[12],"winning":[13],"(#1)":[14],"and":[15,25,59,105],"#10":[16],"submissions":[17,76],"SemEval-2024":[19],"Task":[20],"6:":[21],"Sharedtask":[22],"on":[23,91],"Hallucinations":[24],"Related":[26],"Observable":[27],"Overgeneration":[28],"Mistakes":[29],"(SHROOM)'s":[30],"two":[31,51],"subtasks.This":[32],"task":[33],"challenged":[34],"its":[35],"participants":[36],"to":[37,40],"design":[38],"systems":[39,98],"detect":[41],"hallucinations":[42],"in":[43],"Large":[44,64],"Language":[45,65],"(LLM)":[47],"outputs.Team":[48],"HaRMoNEE":[49],"proposes":[50],"architectures:":[52],"(1)":[53],"fine-tuning":[54,71],"an":[55],"off-the-shelf":[56],"transformer-based":[57],"model":[58],"(2)":[60],"prompt":[61],"tuning":[62],"large-scale":[63],"Models":[66],"(LLMs).One":[67],"submission":[68,82,90],"from":[69,83],"approach":[72,86],"outperformed":[73],"all":[74],"other":[75],"model-aware":[79],"subtask;":[80],"one":[81],"prompt-tuning":[85],"is":[87],"10th-best":[89],"leaderboard":[93],"modelagnostic":[96],"subtask.Our":[97],"also":[99],"include":[100],"pre-processing,":[101],"system-specific":[102],"tuning,":[103],"postprocessing,":[104],"evaluation.":[106]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2024-07-31T00:00:00"}