Matches in SemOpenAlex for { <https://semopenalex.org/work/W3201529751> ?p ?o ?g. }
Showing items 1 to 95 of
95
with 100 items per page.
- W3201529751 abstract "Question answering (QA) models for reading comprehension have been demonstrated to exploit unintended dataset biases such as question–context lexical overlap. This hinders QA models from generalizing to under-represented samples such as questions with low lexical overlap. Question generation (QG), a method for augmenting QA datasets, can be a solution for such performance degradation if QG can properly debias QA datasets. However, we discover that recent neural QG models are biased towards generating questions with high lexical overlap, which can amplify the dataset bias. Moreover, our analysis reveals that data augmentation with these QG models frequently impairs the performance on questions with low lexical overlap, while improving that on questions with high lexical overlap. To address this problem, we use a synonym replacement-based approach to augment questions with low lexical overlap. We demonstrate that the proposed data augmentation approach is simple yet effective to mitigate the degradation problem with only 70k synthetic examples." @default.
- W3201529751 created "2021-09-27" @default.
- W3201529751 creator A5038103607 @default.
- W3201529751 creator A5041062417 @default.
- W3201529751 creator A5078963513 @default.
- W3201529751 date "2021-01-01" @default.
- W3201529751 modified "2023-10-16" @default.
- W3201529751 title "Can Question Generation Debias Question Answering Models? A Case Study on Question–Context Lexical Overlap" @default.
- W3201529751 cites W1522301498 @default.
- W3201529751 cites W1950888772 @default.
- W3201529751 cites W2039987681 @default.
- W3201529751 cites W2081580037 @default.
- W3201529751 cites W2101105183 @default.
- W3201529751 cites W2606333299 @default.
- W3201529751 cites W2889453388 @default.
- W3201529751 cites W2911681509 @default.
- W3201529751 cites W2950733407 @default.
- W3201529751 cites W2951873305 @default.
- W3201529751 cites W2953039212 @default.
- W3201529751 cites W2962717047 @default.
- W3201529751 cites W2962944953 @default.
- W3201529751 cites W2962977247 @default.
- W3201529751 cites W2963108767 @default.
- W3201529751 cites W2963175042 @default.
- W3201529751 cites W2963216553 @default.
- W3201529751 cites W2963341956 @default.
- W3201529751 cites W2963526187 @default.
- W3201529751 cites W2963748441 @default.
- W3201529751 cites W2963938442 @default.
- W3201529751 cites W2963969878 @default.
- W3201529751 cites W2964165364 @default.
- W3201529751 cites W2971296908 @default.
- W3201529751 cites W2980282514 @default.
- W3201529751 cites W3015721781 @default.
- W3201529751 cites W3016970897 @default.
- W3201529751 cites W3032717904 @default.
- W3201529751 cites W3034584102 @default.
- W3201529751 cites W3034716087 @default.
- W3201529751 cites W3103621845 @default.
- W3201529751 doi "https://doi.org/10.18653/v1/2021.mrqa-1.6" @default.
- W3201529751 hasPublicationYear "2021" @default.
- W3201529751 type Work @default.
- W3201529751 sameAs 3201529751 @default.
- W3201529751 citedByCount "1" @default.
- W3201529751 crossrefType "proceedings-article" @default.
- W3201529751 hasAuthorship W3201529751A5038103607 @default.
- W3201529751 hasAuthorship W3201529751A5041062417 @default.
- W3201529751 hasAuthorship W3201529751A5078963513 @default.
- W3201529751 hasBestOaLocation W32015297511 @default.
- W3201529751 hasConcept C126706616 @default.
- W3201529751 hasConcept C151730666 @default.
- W3201529751 hasConcept C154945302 @default.
- W3201529751 hasConcept C157369684 @default.
- W3201529751 hasConcept C165696696 @default.
- W3201529751 hasConcept C173483453 @default.
- W3201529751 hasConcept C204321447 @default.
- W3201529751 hasConcept C2779343474 @default.
- W3201529751 hasConcept C38652104 @default.
- W3201529751 hasConcept C41008148 @default.
- W3201529751 hasConcept C44291984 @default.
- W3201529751 hasConcept C59822182 @default.
- W3201529751 hasConcept C86803240 @default.
- W3201529751 hasConcept C93258239 @default.
- W3201529751 hasConceptScore W3201529751C126706616 @default.
- W3201529751 hasConceptScore W3201529751C151730666 @default.
- W3201529751 hasConceptScore W3201529751C154945302 @default.
- W3201529751 hasConceptScore W3201529751C157369684 @default.
- W3201529751 hasConceptScore W3201529751C165696696 @default.
- W3201529751 hasConceptScore W3201529751C173483453 @default.
- W3201529751 hasConceptScore W3201529751C204321447 @default.
- W3201529751 hasConceptScore W3201529751C2779343474 @default.
- W3201529751 hasConceptScore W3201529751C38652104 @default.
- W3201529751 hasConceptScore W3201529751C41008148 @default.
- W3201529751 hasConceptScore W3201529751C44291984 @default.
- W3201529751 hasConceptScore W3201529751C59822182 @default.
- W3201529751 hasConceptScore W3201529751C86803240 @default.
- W3201529751 hasConceptScore W3201529751C93258239 @default.
- W3201529751 hasLocation W32015297511 @default.
- W3201529751 hasLocation W32015297512 @default.
- W3201529751 hasOpenAccess W3201529751 @default.
- W3201529751 hasPrimaryLocation W32015297511 @default.
- W3201529751 hasRelatedWork W128392744 @default.
- W3201529751 hasRelatedWork W1483367581 @default.
- W3201529751 hasRelatedWork W1518289136 @default.
- W3201529751 hasRelatedWork W1527340856 @default.
- W3201529751 hasRelatedWork W2017258330 @default.
- W3201529751 hasRelatedWork W207304934 @default.
- W3201529751 hasRelatedWork W2747680751 @default.
- W3201529751 hasRelatedWork W3107474891 @default.
- W3201529751 hasRelatedWork W3201529751 @default.
- W3201529751 hasRelatedWork W3213308086 @default.
- W3201529751 isParatext "false" @default.
- W3201529751 isRetracted "false" @default.
- W3201529751 magId "3201529751" @default.
- W3201529751 workType "article" @default.