This project will investigate the use of linguistic features in training a classifier to predict whether comments in online discussions express violent threats. The experiments will be based on a data set comprising 25.000 sentences extracted from YouTube comments. Each sentence has been manually annotated to indicate whether it expresses threats of violence.
Detecting threats of violence in online discussions
Publisert 14. sep. 2014 22:24
- Sist endret 23. sep. 2014 15:03