Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287029227> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4287029227 abstract "Bias mitigation approaches reduce models' dependence on sensitive features of data, such as social group tokens (SGTs), resulting in equal predictions across the sensitive features. In hate speech detection, however, equalizing model predictions may ignore important differences among targeted social groups, as hate speech can contain stereotypical language specific to each SGT. Here, to take the specific language about each SGT into account, we rely on counterfactual fairness and equalize predictions among counterfactuals, generated by changing the SGTs. Our method evaluates the similarity in sentence likelihoods (via pre-trained language models) among counterfactuals, to treat SGTs equally only within interchangeable contexts. By applying logit pairing to equalize outcomes on the restricted set of counterfactuals for each instance, we improve fairness metrics while preserving model performance on hate speech detection." @default.
- W4287029227 created "2022-07-25" @default.
- W4287029227 creator A5000967986 @default.
- W4287029227 creator A5009408707 @default.
- W4287029227 creator A5015311126 @default.
- W4287029227 creator A5061621036 @default.
- W4287029227 creator A5065952016 @default.
- W4287029227 creator A5088094828 @default.
- W4287029227 date "2021-08-03" @default.
- W4287029227 modified "2023-10-16" @default.
- W4287029227 title "Improving Counterfactual Generation for Fair Hate Speech Detection" @default.
- W4287029227 doi "https://doi.org/10.48550/arxiv.2108.01721" @default.
- W4287029227 hasPublicationYear "2021" @default.
- W4287029227 type Work @default.
- W4287029227 citedByCount "0" @default.
- W4287029227 crossrefType "posted-content" @default.
- W4287029227 hasAuthorship W4287029227A5000967986 @default.
- W4287029227 hasAuthorship W4287029227A5009408707 @default.
- W4287029227 hasAuthorship W4287029227A5015311126 @default.
- W4287029227 hasAuthorship W4287029227A5061621036 @default.
- W4287029227 hasAuthorship W4287029227A5065952016 @default.
- W4287029227 hasAuthorship W4287029227A5088094828 @default.
- W4287029227 hasBestOaLocation W42870292271 @default.
- W4287029227 hasConcept C103278499 @default.
- W4287029227 hasConcept C108650721 @default.
- W4287029227 hasConcept C115961682 @default.
- W4287029227 hasConcept C119857082 @default.
- W4287029227 hasConcept C143095724 @default.
- W4287029227 hasConcept C151956035 @default.
- W4287029227 hasConcept C154945302 @default.
- W4287029227 hasConcept C15744967 @default.
- W4287029227 hasConcept C177264268 @default.
- W4287029227 hasConcept C180747234 @default.
- W4287029227 hasConcept C199360897 @default.
- W4287029227 hasConcept C2777530160 @default.
- W4287029227 hasConcept C41008148 @default.
- W4287029227 hasConcept C71889745 @default.
- W4287029227 hasConcept C77805123 @default.
- W4287029227 hasConceptScore W4287029227C103278499 @default.
- W4287029227 hasConceptScore W4287029227C108650721 @default.
- W4287029227 hasConceptScore W4287029227C115961682 @default.
- W4287029227 hasConceptScore W4287029227C119857082 @default.
- W4287029227 hasConceptScore W4287029227C143095724 @default.
- W4287029227 hasConceptScore W4287029227C151956035 @default.
- W4287029227 hasConceptScore W4287029227C154945302 @default.
- W4287029227 hasConceptScore W4287029227C15744967 @default.
- W4287029227 hasConceptScore W4287029227C177264268 @default.
- W4287029227 hasConceptScore W4287029227C180747234 @default.
- W4287029227 hasConceptScore W4287029227C199360897 @default.
- W4287029227 hasConceptScore W4287029227C2777530160 @default.
- W4287029227 hasConceptScore W4287029227C41008148 @default.
- W4287029227 hasConceptScore W4287029227C71889745 @default.
- W4287029227 hasConceptScore W4287029227C77805123 @default.
- W4287029227 hasLocation W42870292271 @default.
- W4287029227 hasOpenAccess W4287029227 @default.
- W4287029227 hasPrimaryLocation W42870292271 @default.
- W4287029227 hasRelatedWork W1533298435 @default.
- W4287029227 hasRelatedWork W2001658247 @default.
- W4287029227 hasRelatedWork W2027436808 @default.
- W4287029227 hasRelatedWork W2129743901 @default.
- W4287029227 hasRelatedWork W2564889467 @default.
- W4287029227 hasRelatedWork W2984875266 @default.
- W4287029227 hasRelatedWork W3118972661 @default.
- W4287029227 hasRelatedWork W4231782964 @default.
- W4287029227 hasRelatedWork W4320724002 @default.
- W4287029227 hasRelatedWork W4321376826 @default.
- W4287029227 isParatext "false" @default.
- W4287029227 isRetracted "false" @default.
- W4287029227 workType "article" @default.