Papers by Allison Earl
Modeling Annotator Disagreement with Demographic-Aware Experts and Synthetic Perspectives (2026.acl-long)
Copied to clipboard
| Challenge: | Recent work treats disagreements as signal, instead of noise, resulting in a single label and marginalizing minoritized perspectives. |
| Approach: | They propose an approach to modeling annotator disagreement in subjective NLP tasks through architectural and data-centric innovations. |
| Outcome: | The proposed model performs competitively across demographic groups and shows strong results on datasets with high disagreement. |