FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3103639864> ?p ?o ?g. }

• W3103639864 abstract "Social biases present in data are often directly reflected in the predictions of models trained on that data. We analyze gender bias in dialogue data, and examine how this bias is not only replicated, but is also amplified in subsequent generative chit-chat dialogue models. We measure gender bias in six existing dialogue datasets before selecting the most biased one, the multi-player text-based fantasy adventure dataset LIGHT, as a testbed for bias mitigation techniques. We consider three techniques to mitigate gender bias: counterfactual data augmentation, targeted data collection, and bias controlled training. We show that our proposed techniques mitigate gender bias by balancing the genderedness of generated dialogue utterances, and find that they are particularly effective in combination. We evaluate model performance with a variety of quantitative methods---including the quantity of gendered words, a dialogue safety classifier, and human assessments---all of which show that our models generate less gendered, but equally engaging chit-chat responses." @default.
• W3103639864 created "2020-11-23" @default.
• W3103639864 creator A5016956470 @default.
• W3103639864 creator A5061347220 @default.
• W3103639864 creator A5062696000 @default.
• W3103639864 creator A5076635608 @default.
• W3103639864 creator A5080241878 @default.
• W3103639864 creator A5083185771 @default.
• W3103639864 date "2020-01-01" @default.
• W3103639864 modified "2023-10-18" @default.
• W3103639864 title "Queens are Powerful too: Mitigating Gender Bias in Dialogue Generation" @default.
• W3103639864 cites W2038466428 @default.
• W3103639864 cites W2057121519 @default.
• W3103639864 cites W2112065548 @default.
• W3103639864 cites W2115261292 @default.
• W3103639864 cites W2129013046 @default.
• W3103639864 cites W2483215953 @default.
• W3103639864 cites W2528130257 @default.
• W3103639864 cites W2573660794 @default.
• W3103639864 cites W2604487042 @default.
• W3103639864 cites W2774008708 @default.
• W3103639864 cites W2799258637 @default.
• W3103639864 cites W2884784896 @default.
• W3103639864 cites W2889624842 @default.
• W3103639864 cites W2890394457 @default.
• W3103639864 cites W2899501643 @default.
• W3103639864 cites W2899513582 @default.
• W3103639864 cites W2908709331 @default.
• W3103639864 cites W2916772188 @default.
• W3103639864 cites W2926555354 @default.
• W3103639864 cites W2941985495 @default.
• W3103639864 cites W2942580297 @default.
• W3103639864 cites W2949744925 @default.
• W3103639864 cites W2949969209 @default.
• W3103639864 cites W2950866572 @default.
• W3103639864 cites W2950888501 @default.
• W3103639864 cites W2952328691 @default.
• W3103639864 cites W2952355715 @default.
• W3103639864 cites W2962787423 @default.
• W3103639864 cites W2962805889 @default.
• W3103639864 cites W2962974452 @default.
• W3103639864 cites W2962990575 @default.
• W3103639864 cites W2963018534 @default.
• W3103639864 cites W2963096510 @default.
• W3103639864 cites W2963206148 @default.
• W3103639864 cites W2963524349 @default.
• W3103639864 cites W2963526187 @default.
• W3103639864 cites W2963723885 @default.
• W3103639864 cites W2963825865 @default.
• W3103639864 cites W2963919731 @default.
• W3103639864 cites W2964352131 @default.
• W3103639864 cites W2970019270 @default.
• W3103639864 cites W2970252402 @default.
• W3103639864 cites W2970387952 @default.
• W3103639864 cites W2970408399 @default.
• W3103639864 cites W2970712718 @default.
• W3103639864 cites W2970723691 @default.
• W3103639864 cites W2970800693 @default.
• W3103639864 cites W2970879410 @default.
• W3103639864 cites W2970948593 @default.
• W3103639864 cites W2971015127 @default.
• W3103639864 cites W2971173235 @default.
• W3103639864 cites W2971307358 @default.
• W3103639864 cites W2971883198 @default.
• W3103639864 cites W2972654913 @default.
• W3103639864 cites W2972668795 @default.
• W3103639864 cites W2979463066 @default.
• W3103639864 cites W2984256198 @default.
• W3103639864 cites W2997607995 @default.
• W3103639864 cites W3027316425 @default.
• W3103639864 cites W3034515982 @default.
• W3103639864 cites W3034778990 @default.
• W3103639864 cites W3034847753 @default.
• W3103639864 cites W3035591180 @default.
• W3103639864 cites W3036997889 @default.
• W3103639864 cites W3037132330 @default.
• W3103639864 cites W3037286488 @default.
• W3103639864 cites W3037378535 @default.
• W3103639864 cites W3037541370 @default.
• W3103639864 cites W3037696302 @default.
• W3103639864 cites W3037697022 @default.
• W3103639864 cites W3037831233 @default.
• W3103639864 cites W3105882417 @default.
• W3103639864 cites W3117655171 @default.
• W3103639864 cites W643597989 @default.
• W3103639864 doi "https://doi.org/10.18653/v1/2020.emnlp-main.656" @default.
• W3103639864 hasPublicationYear "2020" @default.
• W3103639864 type Work @default.
• W3103639864 sameAs 3103639864 @default.
• W3103639864 citedByCount "67" @default.
• W3103639864 countsByYear W31036398642019 @default.
• W3103639864 countsByYear W31036398642020 @default.
• W3103639864 countsByYear W31036398642021 @default.
• W3103639864 countsByYear W31036398642022 @default.
• W3103639864 countsByYear W31036398642023 @default.
• W3103639864 crossrefType "proceedings-article" @default.
• W3103639864 hasAuthorship W3103639864A5016956470 @default.
• W3103639864 hasAuthorship W3103639864A5061347220 @default.
• W3103639864 hasAuthorship W3103639864A5062696000 @default.
• W3103639864 hasAuthorship W3103639864A5076635608 @default.

• next