Matches in SemOpenAlex for { <https://semopenalex.org/work/W3199212893> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W3199212893 abstract "Answering complex open-domain questions requires understanding the latent relations between involving entities. However, we found that the existing QA datasets are extremely imbalanced in some types of relations, which hurts the generalization performance over questions with long-tail relations. To remedy this problem, in this paper, we propose a Relation-Guided Pre-Training (RGPT-QA) framework. We first generate a relational QA dataset covering a wide range of relations from both the Wikidata triplets and Wikipedia hyperlinks. We then pre-train a QA model to infer the latent relations from the question, and then conduct extractive QA to get the target answer entity. We demonstrate that by pretraining with propoed RGPT-QA techique, the popular open-domain QA model, Dense Passage Retriever (DPR), achieves 2.2%, 2.4%, and 6.3% absolute improvement in Exact Match accuracy on Natural Questions, TriviaQA, and WebQuestions. Particularly, we show that RGPT-QA improves significantly on questions with long-tail relations" @default.
- W3199212893 created "2021-09-27" @default.
- W3199212893 creator A5025213473 @default.
- W3199212893 creator A5060639952 @default.
- W3199212893 creator A5087096372 @default.
- W3199212893 date "2021-01-01" @default.
- W3199212893 modified "2023-09-27" @default.
- W3199212893 title "Relation-Guided Pre-Training for Open-Domain Question Answering" @default.
- W3199212893 cites W2080133951 @default.
- W3199212893 cites W2912924812 @default.
- W3199212893 cites W2951434086 @default.
- W3199212893 cites W2953356739 @default.
- W3199212893 cites W2963056065 @default.
- W3199212893 cites W2963175042 @default.
- W3199212893 cites W2963430447 @default.
- W3199212893 cites W2970168256 @default.
- W3199212893 cites W2970986510 @default.
- W3199212893 cites W2989312920 @default.
- W3199212893 cites W2994915912 @default.
- W3199212893 cites W2995636882 @default.
- W3199212893 cites W2995638926 @default.
- W3199212893 cites W3007672467 @default.
- W3199212893 cites W3034671305 @default.
- W3199212893 cites W3034879633 @default.
- W3199212893 cites W3035160371 @default.
- W3199212893 cites W3039578880 @default.
- W3199212893 cites W3082274269 @default.
- W3199212893 cites W3099700870 @default.
- W3199212893 cites W3100859887 @default.
- W3199212893 cites W3101007570 @default.
- W3199212893 cites W3131343200 @default.
- W3199212893 cites W3151929433 @default.
- W3199212893 cites W3167783161 @default.
- W3199212893 doi "https://doi.org/10.18653/v1/2021.findings-emnlp.292" @default.
- W3199212893 hasPublicationYear "2021" @default.
- W3199212893 type Work @default.
- W3199212893 sameAs 3199212893 @default.
- W3199212893 citedByCount "0" @default.
- W3199212893 crossrefType "proceedings-article" @default.
- W3199212893 hasAuthorship W3199212893A5025213473 @default.
- W3199212893 hasAuthorship W3199212893A5060639952 @default.
- W3199212893 hasAuthorship W3199212893A5087096372 @default.
- W3199212893 hasBestOaLocation W31992128931 @default.
- W3199212893 hasConcept C124101348 @default.
- W3199212893 hasConcept C134306372 @default.
- W3199212893 hasConcept C136764020 @default.
- W3199212893 hasConcept C154945302 @default.
- W3199212893 hasConcept C159985019 @default.
- W3199212893 hasConcept C177148314 @default.
- W3199212893 hasConcept C192562407 @default.
- W3199212893 hasConcept C204321447 @default.
- W3199212893 hasConcept C204323151 @default.
- W3199212893 hasConcept C21959979 @default.
- W3199212893 hasConcept C23123220 @default.
- W3199212893 hasConcept C25343380 @default.
- W3199212893 hasConcept C2993776861 @default.
- W3199212893 hasConcept C30088001 @default.
- W3199212893 hasConcept C33923547 @default.
- W3199212893 hasConcept C36503486 @default.
- W3199212893 hasConcept C41008148 @default.
- W3199212893 hasConcept C44291984 @default.
- W3199212893 hasConceptScore W3199212893C124101348 @default.
- W3199212893 hasConceptScore W3199212893C134306372 @default.
- W3199212893 hasConceptScore W3199212893C136764020 @default.
- W3199212893 hasConceptScore W3199212893C154945302 @default.
- W3199212893 hasConceptScore W3199212893C159985019 @default.
- W3199212893 hasConceptScore W3199212893C177148314 @default.
- W3199212893 hasConceptScore W3199212893C192562407 @default.
- W3199212893 hasConceptScore W3199212893C204321447 @default.
- W3199212893 hasConceptScore W3199212893C204323151 @default.
- W3199212893 hasConceptScore W3199212893C21959979 @default.
- W3199212893 hasConceptScore W3199212893C23123220 @default.
- W3199212893 hasConceptScore W3199212893C25343380 @default.
- W3199212893 hasConceptScore W3199212893C2993776861 @default.
- W3199212893 hasConceptScore W3199212893C30088001 @default.
- W3199212893 hasConceptScore W3199212893C33923547 @default.
- W3199212893 hasConceptScore W3199212893C36503486 @default.
- W3199212893 hasConceptScore W3199212893C41008148 @default.
- W3199212893 hasConceptScore W3199212893C44291984 @default.
- W3199212893 hasLocation W31992128931 @default.
- W3199212893 hasLocation W31992128932 @default.
- W3199212893 hasOpenAccess W3199212893 @default.
- W3199212893 hasPrimaryLocation W31992128931 @default.
- W3199212893 hasRelatedWork W11133913 @default.
- W3199212893 hasRelatedWork W11196620 @default.
- W3199212893 hasRelatedWork W14310100 @default.
- W3199212893 hasRelatedWork W14427702 @default.
- W3199212893 hasRelatedWork W155019 @default.
- W3199212893 hasRelatedWork W4052871 @default.
- W3199212893 hasRelatedWork W5516682 @default.
- W3199212893 hasRelatedWork W5772686 @default.
- W3199212893 hasRelatedWork W6865571 @default.
- W3199212893 hasRelatedWork W10597021 @default.
- W3199212893 isParatext "false" @default.
- W3199212893 isRetracted "false" @default.
- W3199212893 magId "3199212893" @default.
- W3199212893 workType "article" @default.