Matches in SemOpenAlex for { <https://semopenalex.org/work/W4362529236> ?p ?o ?g. }
- W4362529236 endingPage "103367" @default.
- W4362529236 startingPage "103367" @default.
- W4362529236 abstract "Many existing biomedical extractive question answering methods are based on pre-trained models, which do not take full advantage of the hidden layer knowledge of pretrained models and do not consider span overlap between answers when predicting. To address these issues, we propose a new question answering model, called ALBERT with Dynamic Routing and Answer Voting (ADRAV). The ADRAV can reasonably utilize hidden layer knowledge through dynamic routing, and consider span similarity between answers through answer voting. To improve the performance of the model, we also carry out pre-fine-tuning, and add a dynamic parameter adjustment mechanism in the process of pre-fine-tuning. Experimental results show that our model achieves significant performance improvement with fewer parameters on BioASQ 4b, 5b, 6b, 9b, and outperforms SOTA baselines on BioASQ 4b, 6b." @default.
- W4362529236 created "2023-04-06" @default.
- W4362529236 creator A5025314679 @default.
- W4362529236 creator A5029178260 @default.
- W4362529236 creator A5036530520 @default.
- W4362529236 creator A5048709513 @default.
- W4362529236 creator A5083581319 @default.
- W4362529236 date "2023-07-01" @default.
- W4362529236 modified "2023-09-26" @default.
- W4362529236 title "Biomedical extractive question answering based on dynamic routing and answer voting" @default.
- W4362529236 cites W1981208470 @default.
- W4362529236 cites W2740815822 @default.
- W4362529236 cites W2886993012 @default.
- W4362529236 cites W2906902434 @default.
- W4362529236 cites W2911489562 @default.
- W4362529236 cites W2919358988 @default.
- W4362529236 cites W2947411064 @default.
- W4362529236 cites W2966359917 @default.
- W4362529236 cites W2979826702 @default.
- W4362529236 cites W2980708516 @default.
- W4362529236 cites W2982902390 @default.
- W4362529236 cites W2990887822 @default.
- W4362529236 cites W2997428102 @default.
- W4362529236 cites W3001393026 @default.
- W4362529236 cites W3008191852 @default.
- W4362529236 cites W3011574394 @default.
- W4362529236 cites W3013838212 @default.
- W4362529236 cites W3024333932 @default.
- W4362529236 cites W3035763680 @default.
- W4362529236 cites W3086388855 @default.
- W4362529236 cites W3115034342 @default.
- W4362529236 cites W3127542694 @default.
- W4362529236 cites W3174522966 @default.
- W4362529236 cites W3179963059 @default.
- W4362529236 cites W3196227100 @default.
- W4362529236 cites W3197896619 @default.
- W4362529236 cites W3198080531 @default.
- W4362529236 cites W3199972011 @default.
- W4362529236 cites W3202548348 @default.
- W4362529236 cites W3202929920 @default.
- W4362529236 cites W3212275514 @default.
- W4362529236 cites W4200150102 @default.
- W4362529236 cites W4200442472 @default.
- W4362529236 cites W4226172857 @default.
- W4362529236 cites W4281394921 @default.
- W4362529236 doi "https://doi.org/10.1016/j.ipm.2023.103367" @default.
- W4362529236 hasPublicationYear "2023" @default.
- W4362529236 type Work @default.
- W4362529236 citedByCount "0" @default.
- W4362529236 crossrefType "journal-article" @default.
- W4362529236 hasAuthorship W4362529236A5025314679 @default.
- W4362529236 hasAuthorship W4362529236A5029178260 @default.
- W4362529236 hasAuthorship W4362529236A5036530520 @default.
- W4362529236 hasAuthorship W4362529236A5048709513 @default.
- W4362529236 hasAuthorship W4362529236A5083581319 @default.
- W4362529236 hasConcept C10138342 @default.
- W4362529236 hasConcept C103278499 @default.
- W4362529236 hasConcept C104954878 @default.
- W4362529236 hasConcept C111919701 @default.
- W4362529236 hasConcept C115961682 @default.
- W4362529236 hasConcept C119857082 @default.
- W4362529236 hasConcept C154945302 @default.
- W4362529236 hasConcept C162324750 @default.
- W4362529236 hasConcept C17744445 @default.
- W4362529236 hasConcept C178790620 @default.
- W4362529236 hasConcept C185592680 @default.
- W4362529236 hasConcept C199539241 @default.
- W4362529236 hasConcept C204948658 @default.
- W4362529236 hasConcept C24856439 @default.
- W4362529236 hasConcept C2776299755 @default.
- W4362529236 hasConcept C2779227376 @default.
- W4362529236 hasConcept C31258907 @default.
- W4362529236 hasConcept C41008148 @default.
- W4362529236 hasConcept C44291984 @default.
- W4362529236 hasConcept C520049643 @default.
- W4362529236 hasConcept C74172769 @default.
- W4362529236 hasConcept C94625758 @default.
- W4362529236 hasConcept C98045186 @default.
- W4362529236 hasConceptScore W4362529236C10138342 @default.
- W4362529236 hasConceptScore W4362529236C103278499 @default.
- W4362529236 hasConceptScore W4362529236C104954878 @default.
- W4362529236 hasConceptScore W4362529236C111919701 @default.
- W4362529236 hasConceptScore W4362529236C115961682 @default.
- W4362529236 hasConceptScore W4362529236C119857082 @default.
- W4362529236 hasConceptScore W4362529236C154945302 @default.
- W4362529236 hasConceptScore W4362529236C162324750 @default.
- W4362529236 hasConceptScore W4362529236C17744445 @default.
- W4362529236 hasConceptScore W4362529236C178790620 @default.
- W4362529236 hasConceptScore W4362529236C185592680 @default.
- W4362529236 hasConceptScore W4362529236C199539241 @default.
- W4362529236 hasConceptScore W4362529236C204948658 @default.
- W4362529236 hasConceptScore W4362529236C24856439 @default.
- W4362529236 hasConceptScore W4362529236C2776299755 @default.
- W4362529236 hasConceptScore W4362529236C2779227376 @default.
- W4362529236 hasConceptScore W4362529236C31258907 @default.
- W4362529236 hasConceptScore W4362529236C41008148 @default.
- W4362529236 hasConceptScore W4362529236C44291984 @default.
- W4362529236 hasConceptScore W4362529236C520049643 @default.