FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3192500523> ?p ?o ?g. }

• W3192500523 abstract "Obtaining human-like performance in NLP is often argued to require compositional generalisation. Whether neural networks exhibit this ability is usually studied by training models on highly compositional synthetic data. However, compositionality in natural language is much more complex than the rigid, arithmetic-like version such data adheres to, and artificial compositionality tests thus do not allow us to determine how neural models deal with more realistic forms of compositionality. In this work, we re-instantiate three compositionality tests from the literature and reformulate them for neural machine translation (NMT).Our results highlight that: i) unfavourably, models trained on more data are more compositional; ii) models are sometimes less compositional than expected, but sometimes more, exemplifying that different levels of compositionality are required, and models are not always able to modulate between them correctly; iii) some of the non-compositional behaviours are mistakes, whereas others reflect the natural variation in data. Apart from an empirical study, our work is a call to action: we should rethink the evaluation of compositionality in neural networks and develop benchmarks using real data to evaluate compositionality on natural language, where composing meaning is not as straightforward as doing the math." @default.
• W3192500523 created "2021-08-16" @default.
• W3192500523 creator A5007465102 @default.
• W3192500523 creator A5018069499 @default.
• W3192500523 creator A5074102828 @default.
• W3192500523 date "2022-01-01" @default.
• W3192500523 modified "2023-10-14" @default.
• W3192500523 title "The Paradox of the Compositionality of Natural Language: A Neural Machine Translation Case Study" @default.
• W3192500523 cites W1699946128 @default.
• W3192500523 cites W2013494846 @default.
• W3192500523 cites W2016954686 @default.
• W3192500523 cites W2067041071 @default.
• W3192500523 cites W2118373646 @default.
• W3192500523 cites W2165266220 @default.
• W3192500523 cites W2336011916 @default.
• W3192500523 cites W2515796212 @default.
• W3192500523 cites W2571532437 @default.
• W3192500523 cites W2622377691 @default.
• W3192500523 cites W2743630524 @default.
• W3192500523 cites W2798663534 @default.
• W3192500523 cites W2922523190 @default.
• W3192500523 cites W2933138175 @default.
• W3192500523 cites W2947029787 @default.
• W3192500523 cites W2951286828 @default.
• W3192500523 cites W2963267799 @default.
• W3192500523 cites W2963403868 @default.
• W3192500523 cites W2972433021 @default.
• W3192500523 cites W2978165923 @default.
• W3192500523 cites W2986378306 @default.
• W3192500523 cites W2986889180 @default.
• W3192500523 cites W2990379018 @default.
• W3192500523 cites W3007855821 @default.
• W3192500523 cites W3017374003 @default.
• W3192500523 cites W3030374439 @default.
• W3192500523 cites W3035331128 @default.
• W3192500523 cites W3035473547 @default.
• W3192500523 cites W3045885009 @default.
• W3192500523 cites W3082928416 @default.
• W3192500523 cites W3092580630 @default.
• W3192500523 cites W3104739822 @default.
• W3192500523 cites W3119000810 @default.
• W3192500523 cites W3152540792 @default.
• W3192500523 cites W3162810836 @default.
• W3192500523 cites W3169369929 @default.
• W3192500523 cites W3176846012 @default.
• W3192500523 cites W3177804148 @default.
• W3192500523 doi "https://doi.org/10.18653/v1/2022.acl-long.286" @default.
• W3192500523 hasPublicationYear "2022" @default.
• W3192500523 type Work @default.
• W3192500523 sameAs 3192500523 @default.
• W3192500523 citedByCount "5" @default.
• W3192500523 countsByYear W31925005232021 @default.
• W3192500523 countsByYear W31925005232022 @default.
• W3192500523 crossrefType "proceedings-article" @default.
• W3192500523 hasAuthorship W3192500523A5007465102 @default.
• W3192500523 hasAuthorship W3192500523A5018069499 @default.
• W3192500523 hasAuthorship W3192500523A5074102828 @default.
• W3192500523 hasBestOaLocation W31925005231 @default.
• W3192500523 hasConcept C104317684 @default.
• W3192500523 hasConcept C105580179 @default.
• W3192500523 hasConcept C121375916 @default.
• W3192500523 hasConcept C149364088 @default.
• W3192500523 hasConcept C154945302 @default.
• W3192500523 hasConcept C166957645 @default.
• W3192500523 hasConcept C185592680 @default.
• W3192500523 hasConcept C195324797 @default.
• W3192500523 hasConcept C203005215 @default.
• W3192500523 hasConcept C204321447 @default.
• W3192500523 hasConcept C2776608160 @default.
• W3192500523 hasConcept C41008148 @default.
• W3192500523 hasConcept C50644808 @default.
• W3192500523 hasConcept C55493867 @default.
• W3192500523 hasConcept C95457728 @default.
• W3192500523 hasConceptScore W3192500523C104317684 @default.
• W3192500523 hasConceptScore W3192500523C105580179 @default.
• W3192500523 hasConceptScore W3192500523C121375916 @default.
• W3192500523 hasConceptScore W3192500523C149364088 @default.
• W3192500523 hasConceptScore W3192500523C154945302 @default.
• W3192500523 hasConceptScore W3192500523C166957645 @default.
• W3192500523 hasConceptScore W3192500523C185592680 @default.
• W3192500523 hasConceptScore W3192500523C195324797 @default.
• W3192500523 hasConceptScore W3192500523C203005215 @default.
• W3192500523 hasConceptScore W3192500523C204321447 @default.
• W3192500523 hasConceptScore W3192500523C2776608160 @default.
• W3192500523 hasConceptScore W3192500523C41008148 @default.
• W3192500523 hasConceptScore W3192500523C50644808 @default.
• W3192500523 hasConceptScore W3192500523C55493867 @default.
• W3192500523 hasConceptScore W3192500523C95457728 @default.
• W3192500523 hasLocation W31925005231 @default.
• W3192500523 hasLocation W31925005232 @default.
• W3192500523 hasOpenAccess W3192500523 @default.
• W3192500523 hasPrimaryLocation W31925005231 @default.
• W3192500523 hasRelatedWork W1484029852 @default.
• W3192500523 hasRelatedWork W1512718085 @default.
• W3192500523 hasRelatedWork W1569841287 @default.
• W3192500523 hasRelatedWork W193726211 @default.
• W3192500523 hasRelatedWork W1978161643 @default.
• W3192500523 hasRelatedWork W2140460368 @default.
• W3192500523 hasRelatedWork W2167662847 @default.
• W3192500523 hasRelatedWork W3198474835 @default.

• next