Papers with Natural Language Processing models
“It’s Not Just Hate”: A Multi-Dimensional Perspective on Detecting Harmful Speech Online (2022.emnlp-main)
Copied to clipboard
| Challenge: | Detecting offensive content is becoming a critical task in natural language processing . but most datasets use a single binary label for hate or incivility, even though each concept is multi-faceted . a more fine-grained multi-label approach addresses conceptual and performance issues . |
| Approach: | They propose to use a dataset to annotate offensive online speech with six labels . they propose to apply a more fine-grained approach to predicting incivility and hateful content . |
| Outcome: | The proposed approach outperforms or matches benchmark datasets on the annotated tweets. |
ActiveAED: A Human in the Loop Improves Annotation Error Detection (2023.findings-acl)
Copied to clipboard
| Challenge: | Annotation error detection (AED) methods flag erroneous annotations for human reannotation . but many of these methods assume a final curation step in which a human annotator decides whether the annotation is erranosome . |
| Approach: | They propose an annotation error detection method that can detect errors more accurately by repeatedly querying a human for error corrections in its prediction loop. |
| Outcome: | The proposed method improves on eight datasets spanning five different tasks and can improve accuracy by up to six percentage points. |