Matches in SemOpenAlex for { <https://semopenalex.org/work/W4306888994> ?p ?o ?g. }
- W4306888994 endingPage "35" @default.
- W4306888994 startingPage "27" @default.
- W4306888994 abstract "Pre-trained language models (PrLMs) have achieved great success on a wide range of natural language processing tasks by virtue of the universal language representation ability obtained by self-supervised learning on a large corpus. These models are pre-trained on standard plain texts with general language model (LM) training objectives, which would be insufficient to model dialogue-exclusive attributes like specificity and informativeness reflected in these tasks that are not explicitly captured by the pre-trained universal language representations. In this work, we propose dialogue-adaptive pre-training objectives (DAPO) derived from quality estimation to simulate dialogue-specific features, namely coherence, specificity, and informativeness. As the foundation for model pre-training, we synthesize a new dialogue corpus and build our training set with two unsupervised methods: 1) coherence-oriented context corruption, including utterance ordering, insertion, and replacement, to help the model capture the coherence inside the dialogue contexts; and 2) specificity-oriented automatic rescoring, which encourages the model to measure the quality of the synthesized data for dialogue-adaptive pre-training by considering specificity and informativeness. Experimental results on widely used open-domain response selection and quality estimation benchmarks show that DAPO significantly improves the baseline models and achieves state-of-the-art performance on the MuTual leaderboard, verifying the effectiveness of estimating quality evaluation factors into pre-training." @default.
- W4306888994 created "2022-10-21" @default.
- W4306888994 creator A5036050911 @default.
- W4306888994 creator A5070962435 @default.
- W4306888994 creator A5077320577 @default.
- W4306888994 date "2023-01-01" @default.
- W4306888994 modified "2023-10-06" @default.
- W4306888994 title "Dialogue-adaptive language model pre-training from quality estimation☆" @default.
- W4306888994 cites W2095652037 @default.
- W4306888994 cites W2101105183 @default.
- W4306888994 cites W2270070752 @default.
- W4306888994 cites W2798456655 @default.
- W4306888994 cites W2891416139 @default.
- W4306888994 cites W2916772188 @default.
- W4306888994 cites W2962854379 @default.
- W4306888994 cites W2963341956 @default.
- W4306888994 cites W2963527228 @default.
- W4306888994 cites W2963825865 @default.
- W4306888994 cites W2963866450 @default.
- W4306888994 cites W2972664115 @default.
- W4306888994 cites W2988937804 @default.
- W4306888994 cites W2990563493 @default.
- W4306888994 cites W2996876695 @default.
- W4306888994 cites W2997200074 @default.
- W4306888994 cites W3034446185 @default.
- W4306888994 cites W3034600233 @default.
- W4306888994 cites W3034808773 @default.
- W4306888994 cites W3034927876 @default.
- W4306888994 cites W3034950505 @default.
- W4306888994 cites W3035451444 @default.
- W4306888994 cites W3086456409 @default.
- W4306888994 cites W3097517997 @default.
- W4306888994 cites W3104078590 @default.
- W4306888994 cites W3113677131 @default.
- W4306888994 cites W3129125493 @default.
- W4306888994 cites W3133407825 @default.
- W4306888994 cites W3157689701 @default.
- W4306888994 cites W3168988646 @default.
- W4306888994 doi "https://doi.org/10.1016/j.neucom.2022.10.036" @default.
- W4306888994 hasPublicationYear "2023" @default.
- W4306888994 type Work @default.
- W4306888994 citedByCount "1" @default.
- W4306888994 countsByYear W43068889942023 @default.
- W4306888994 crossrefType "journal-article" @default.
- W4306888994 hasAuthorship W4306888994A5036050911 @default.
- W4306888994 hasAuthorship W4306888994A5070962435 @default.
- W4306888994 hasAuthorship W4306888994A5077320577 @default.
- W4306888994 hasBestOaLocation W43068889942 @default.
- W4306888994 hasConcept C111472728 @default.
- W4306888994 hasConcept C119857082 @default.
- W4306888994 hasConcept C121332964 @default.
- W4306888994 hasConcept C134752490 @default.
- W4306888994 hasConcept C137293760 @default.
- W4306888994 hasConcept C138885662 @default.
- W4306888994 hasConcept C151730666 @default.
- W4306888994 hasConcept C154945302 @default.
- W4306888994 hasConcept C177264268 @default.
- W4306888994 hasConcept C17744445 @default.
- W4306888994 hasConcept C199360897 @default.
- W4306888994 hasConcept C199539241 @default.
- W4306888994 hasConcept C204321447 @default.
- W4306888994 hasConcept C2775852435 @default.
- W4306888994 hasConcept C2776359362 @default.
- W4306888994 hasConcept C2779343474 @default.
- W4306888994 hasConcept C2779530757 @default.
- W4306888994 hasConcept C2781181686 @default.
- W4306888994 hasConcept C41008148 @default.
- W4306888994 hasConcept C62520636 @default.
- W4306888994 hasConcept C86803240 @default.
- W4306888994 hasConcept C94625758 @default.
- W4306888994 hasConcept C95318506 @default.
- W4306888994 hasConceptScore W4306888994C111472728 @default.
- W4306888994 hasConceptScore W4306888994C119857082 @default.
- W4306888994 hasConceptScore W4306888994C121332964 @default.
- W4306888994 hasConceptScore W4306888994C134752490 @default.
- W4306888994 hasConceptScore W4306888994C137293760 @default.
- W4306888994 hasConceptScore W4306888994C138885662 @default.
- W4306888994 hasConceptScore W4306888994C151730666 @default.
- W4306888994 hasConceptScore W4306888994C154945302 @default.
- W4306888994 hasConceptScore W4306888994C177264268 @default.
- W4306888994 hasConceptScore W4306888994C17744445 @default.
- W4306888994 hasConceptScore W4306888994C199360897 @default.
- W4306888994 hasConceptScore W4306888994C199539241 @default.
- W4306888994 hasConceptScore W4306888994C204321447 @default.
- W4306888994 hasConceptScore W4306888994C2775852435 @default.
- W4306888994 hasConceptScore W4306888994C2776359362 @default.
- W4306888994 hasConceptScore W4306888994C2779343474 @default.
- W4306888994 hasConceptScore W4306888994C2779530757 @default.
- W4306888994 hasConceptScore W4306888994C2781181686 @default.
- W4306888994 hasConceptScore W4306888994C41008148 @default.
- W4306888994 hasConceptScore W4306888994C62520636 @default.
- W4306888994 hasConceptScore W4306888994C86803240 @default.
- W4306888994 hasConceptScore W4306888994C94625758 @default.
- W4306888994 hasConceptScore W4306888994C95318506 @default.
- W4306888994 hasFunder F4320321001 @default.
- W4306888994 hasLocation W43068889941 @default.
- W4306888994 hasLocation W43068889942 @default.
- W4306888994 hasOpenAccess W4306888994 @default.