{"id":"https://openalex.org/W7140107759","doi":"https://doi.org/10.18653/v1/2026.eacl-long.130","title":"PTEB: Towards Robust Text Embedding Evaluation via Stochastic Paraphrasing at Evaluation Time with LLMs","display_name":"PTEB: Towards Robust Text Embedding Evaluation via Stochastic Paraphrasing at Evaluation Time with LLMs","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140107759","doi":"https://doi.org/10.18653/v1/2026.eacl-long.130"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.eacl-long.130","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.130","pdf_url":"https://aclanthology.org/2026.eacl-long.130.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":"Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.eacl-long.130.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Manuel Frank","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Manuel Frank","raw_affiliation_strings":["Department of Computer Science Munster Technological University Department of Computer Science Munster Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science Munster Technological University Department of Computer Science Munster Technological University","institution_ids":[]}]},{"author_position":"last","author":{"id":null,"display_name":"Haithem Afli","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haithem Afli","raw_affiliation_strings":["Department of Computer Science Munster Technological University Department of Computer Science Munster Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science Munster Technological University Department of Computer Science Munster Technological University","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":2,"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":0.3964668,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2832","last_page":"2851"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.660099983215332,"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":0.660099983215332,"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":0.1987999975681305,"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/T13629","display_name":"Text Readability and Simplification","score":0.03660000115633011,"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/embedding","display_name":"Embedding","score":0.4221000075340271},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.3086000084877014},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.2912999987602234},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.2703000009059906}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5893999934196472},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5440000295639038},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.4221000075340271},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3874000012874603},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3246999979019165},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3086000084877014},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.2912999987602234},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2736999988555908},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2703000009059906},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.26089999079704285}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.eacl-long.130","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.130","pdf_url":"https://aclanthology.org/2026.eacl-long.130.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":"Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.eacl-long.130","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.130","pdf_url":"https://aclanthology.org/2026.eacl-long.130.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":"Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1874226381","display_name":null,"funder_award_id":"13/RC/2106_P2","funder_id":"https://openalex.org/F4320336201","funder_display_name":"ADAPT - Centre for Digital Content Technology"},{"id":"https://openalex.org/G3239979300","display_name":null,"funder_award_id":"13/RC/2106_P2","funder_id":"https://openalex.org/F4320335322","funder_display_name":"European Regional Development Fund"}],"funders":[{"id":"https://openalex.org/F4320331797","display_name":"Munster Technological University","ror":null},{"id":"https://openalex.org/F4320334322","display_name":"HORIZON EUROPE Framework Programme","ror":null},{"id":"https://openalex.org/F4320335322","display_name":"European Regional Development Fund","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320336201","display_name":"ADAPT - Centre for Digital Content Technology","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140107759.pdf","grobid_xml":"https://content.openalex.org/works/W7140107759.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Current":[0],"sentence":[1,88],"embedding":[2],"evaluations":[3],"typically":[4],"rely":[5],"on":[6,20,142],"static":[7],"test":[8],"beds":[9],"like":[10],"the":[11,33,85,156],"Massive":[12],"Text":[13,35],"Embedding":[14,36],"Benchmark":[15,37],"(MTEB).While":[16],"invaluable,":[17],"repeated":[18],"tuning":[19],"a":[21,39,56,133],"fixed":[22],"suite":[23],"can":[24],"inflate":[25],"reported":[26],"scores":[27],"and":[28,50,64,125],"obscure":[29],"real-world":[30],"robustness.We":[31],"introduce":[32],"Paraphrasing":[34],"(PTEB),":[38],"dynamic":[40],"protocol":[41],"that":[42,69,84,104,139],"stochastically":[43],"generates":[44],"meaningpreserving":[45],"paraphrases":[46],"at":[47],"evaluation":[48,135,151],"time":[49],"aggregates":[51],"results":[52,115],"across":[53],"multiple":[54,120],"runs.Using":[55],"cost-efficient":[57],"LLM-based":[58],"method":[59],"grounded":[60],"in":[61,94,137],"gold":[62],"ratings":[63],"human":[65],"validation,":[66],"we":[67,80,129],"show":[68],"LLMs":[70],"generate":[71],"token-diverse":[72],"but":[73,146],"semantically":[74],"preserving":[75],"paraphrases.Across":[76],"7":[77],"MTEB":[78],"tasks,":[79],"validate":[81],"our":[82],"hypothesis":[83],"performance":[86],"of":[87],"encoders":[89],"is":[90],"sensitive":[91],"to":[92,112,131,158],"changes":[93],"token":[95],"space":[96],"even":[97],"when":[98],"semantics":[99],"remain":[100],"fixed.We":[101],"also":[102],"observe":[103],"smaller":[105],"models":[106],"are":[107,116],"not":[108],"disproportionately":[109],"affected":[110],"relative":[111],"larger":[113],"ones.Our":[114],"statistically":[117],"robust":[118],"over":[119],"runs":[121],"spanning":[122],"20":[123],"datasets":[124],"25":[126],"languages.More":[127],"generally,":[128],"aim":[130],"propose":[132],"new":[134],"paradigm":[136],"NLP":[138],"relies":[140],"less":[141],"static,":[143],"pre-defined":[144],"benchmarks":[145],"shifts":[147],"towards":[148],"dynamic,":[149],"stochastic":[150],"leveraging":[152],"eval-time":[153],"compute.We":[154],"make":[155],"code":[157],"run":[159],"PTEB":[160],"publicly":[161],"available.":[162],"1":[163]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-03-04T00:00:00"}