Papers by Einat Minkov
A Closer Look at Multidimensional Online Political Incivility (2024.emnlp-main)
Copied to clipboard
| Challenge: | 80% of the uncivil tweets are authored by 20% of the users, where users who are politically engaged are more inclined to use uncival language. |
| Approach: | They analysed 13K political tweets in the U.S. using crowd sourcing and classified them by their respective categories. |
| Outcome: | The proposed method enables us to characterise the distribution of incivility across users and geopolitical regions. |
Fight Fire with Fire: Fine-tuning Hate Detectors using Large Samples of Generated Hate Speech (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for hate speech detection are limited in size and lack of labeled datasets. |
| Approach: | They employ pretrained language models to generate large amounts of hate speech sequences from available labeled examples. |
| Outcome: | The proposed model improves generalization significantly and consistently within and across data distributions. |
Towards Author-informed NLP: Mind the Social Bias (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing models of text understanding fail when opinions are conveyed implicitly or sarcastically. |
| Approach: | They propose to model user contexts within a social embedding space that was learned from the Twitter network at large-scale. |
| Outcome: | The proposed model improves generalization of stance prediction and toxicity detection, and also toxicity and incivility detection. |