| Download | - View final version: Ruddit: norms of offensiveness for English Reddit comments (PDF, 1.2 MiB)
|
|---|
| DOI | Resolve DOI: https://doi.org/10.18653/v1/2021.acl-long.210 |
|---|
| Author | Search for: Hada, Rishav; Search for: Sudhir, Sohi; Search for: Mishra, Pushkar; Search for: Yannakoudakis, Helen; Search for: Mohammad, Saif M.1; Search for: Shutova, Ekaterina |
|---|
| Affiliation | - National Research Council of Canada. Digital Technologies
|
|---|
| Format | Text, Article |
|---|
| Conference | ACL-IJCNLP 2021 - 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), Aug. 1-6, 2021, Virtual Conference |
|---|
| Abstract | On social media platforms, hateful and offensive language negatively impact the mental well-being of users and the participation of people from diverse backgrounds. Automatic methods to detect offensive language have largely relied on datasets with categorical labels. However, comments can vary in their degree of offensiveness. We create the first dataset of English language Reddit comments that has fine-grained, real-valued scores between -1 (maximally supportive) and 1 (maximally offensive). The dataset was annotated using Best–Worst Scaling, a form of comparative annotation that has been shown to alleviate known biases of using rating scales. We show that the method produces highly reliable offensiveness scores. Finally, we evaluate the ability of widely-used neural models to predict offensiveness scores on this new dataset. |
|---|
| Publication date | 2021-08-01 |
|---|
| Publisher | Association for Computational Linguistics |
|---|
| Licence | |
|---|
| In | |
|---|
| Related data | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 9afba8cc-72c9-4f25-a6e1-7e834752c3f4 |
|---|
| Record created | 2021-11-15 |
|---|
| Record modified | 2021-11-16 |
|---|
