Matches in SemOpenAlex for { <https://semopenalex.org/work/W4221165966> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4221165966 abstract "Unlike classical lexical overlap metrics such as BLEU, most current evaluation metrics (such as BERTScore or MoverScore) are based on black-box language models such as BERT or XLM-R. They often achieve strong correlations with human judgments, but recent research indicates that the lower-quality classical metrics remain dominant, one of the potential reasons being that their decision processes are transparent. To foster more widespread acceptance of the novel high-quality metrics, explainability thus becomes crucial. In this concept paper, we identify key properties and propose key goals of explainable machine translation evaluation metrics. We also provide a synthesizing overview over recent approaches for explainable machine translation metrics and discuss how they relate to those goals and properties. Further, we conduct own novel experiments, which (among others) find that current adversarial NLP techniques are unsuitable for automatically identifying limitations of high-quality black-box evaluation metrics, as they are not meaning-preserving. Finally, we provide a vision of future approaches to explainable evaluation metrics and their evaluation. We hope that our work can help catalyze and guide future research on explainable evaluation metrics and, mediately, also contribute to better and more transparent text generation systems." @default.
- W4221165966 created "2022-04-03" @default.
- W4221165966 creator A5004393324 @default.
- W4221165966 creator A5021180431 @default.
- W4221165966 creator A5030347098 @default.
- W4221165966 creator A5040392598 @default.
- W4221165966 creator A5053947568 @default.
- W4221165966 creator A5072097629 @default.
- W4221165966 date "2022-03-21" @default.
- W4221165966 modified "2023-09-25" @default.
- W4221165966 title "Towards Explainable Evaluation Metrics for Natural Language Generation" @default.
- W4221165966 doi "https://doi.org/10.48550/arxiv.2203.11131" @default.
- W4221165966 hasPublicationYear "2022" @default.
- W4221165966 type Work @default.
- W4221165966 citedByCount "0" @default.
- W4221165966 crossrefType "posted-content" @default.
- W4221165966 hasAuthorship W4221165966A5004393324 @default.
- W4221165966 hasAuthorship W4221165966A5021180431 @default.
- W4221165966 hasAuthorship W4221165966A5030347098 @default.
- W4221165966 hasAuthorship W4221165966A5040392598 @default.
- W4221165966 hasAuthorship W4221165966A5053947568 @default.
- W4221165966 hasAuthorship W4221165966A5072097629 @default.
- W4221165966 hasBestOaLocation W42211659661 @default.
- W4221165966 hasConcept C111472728 @default.
- W4221165966 hasConcept C119857082 @default.
- W4221165966 hasConcept C138885662 @default.
- W4221165966 hasConcept C154945302 @default.
- W4221165966 hasConcept C15744967 @default.
- W4221165966 hasConcept C195324797 @default.
- W4221165966 hasConcept C203005215 @default.
- W4221165966 hasConcept C204321447 @default.
- W4221165966 hasConcept C2522767166 @default.
- W4221165966 hasConcept C26517878 @default.
- W4221165966 hasConcept C2779439875 @default.
- W4221165966 hasConcept C2779530757 @default.
- W4221165966 hasConcept C2780876879 @default.
- W4221165966 hasConcept C37736160 @default.
- W4221165966 hasConcept C38652104 @default.
- W4221165966 hasConcept C41008148 @default.
- W4221165966 hasConcept C542102704 @default.
- W4221165966 hasConcept C94966114 @default.
- W4221165966 hasConceptScore W4221165966C111472728 @default.
- W4221165966 hasConceptScore W4221165966C119857082 @default.
- W4221165966 hasConceptScore W4221165966C138885662 @default.
- W4221165966 hasConceptScore W4221165966C154945302 @default.
- W4221165966 hasConceptScore W4221165966C15744967 @default.
- W4221165966 hasConceptScore W4221165966C195324797 @default.
- W4221165966 hasConceptScore W4221165966C203005215 @default.
- W4221165966 hasConceptScore W4221165966C204321447 @default.
- W4221165966 hasConceptScore W4221165966C2522767166 @default.
- W4221165966 hasConceptScore W4221165966C26517878 @default.
- W4221165966 hasConceptScore W4221165966C2779439875 @default.
- W4221165966 hasConceptScore W4221165966C2779530757 @default.
- W4221165966 hasConceptScore W4221165966C2780876879 @default.
- W4221165966 hasConceptScore W4221165966C37736160 @default.
- W4221165966 hasConceptScore W4221165966C38652104 @default.
- W4221165966 hasConceptScore W4221165966C41008148 @default.
- W4221165966 hasConceptScore W4221165966C542102704 @default.
- W4221165966 hasConceptScore W4221165966C94966114 @default.
- W4221165966 hasLocation W42211659661 @default.
- W4221165966 hasOpenAccess W4221165966 @default.
- W4221165966 hasPrimaryLocation W42211659661 @default.
- W4221165966 hasRelatedWork W1538473846 @default.
- W4221165966 hasRelatedWork W1563618553 @default.
- W4221165966 hasRelatedWork W1806995473 @default.
- W4221165966 hasRelatedWork W2368651715 @default.
- W4221165966 hasRelatedWork W2977842567 @default.
- W4221165966 hasRelatedWork W3107474891 @default.
- W4221165966 hasRelatedWork W3118032223 @default.
- W4221165966 hasRelatedWork W3173996958 @default.
- W4221165966 hasRelatedWork W3198474835 @default.
- W4221165966 hasRelatedWork W4287548963 @default.
- W4221165966 isParatext "false" @default.
- W4221165966 isRetracted "false" @default.
- W4221165966 workType "article" @default.