Skip to main content
zenodoopen

nlp2023_toxic_german

<p>The dataset was sourced from the Austrian newspaper DerStandard (www.derStandard.at) and provided in JSON file format. The dataset encompasses responses to articles published between November 4, 2021, and November 10, 2021 and consisted of 4473 postings belonging to 522 different articles. Only original comments were selected and responses to other comments were excluded. With 2818 nontoxic and 1655 toxic entries the dataset is not very unbalanced. Beside the comments and the binary class label, the date was partly also annotated with four different labels with the position in the text.</p> <p>These labels are</p> <ul> <li>Vulgarity Obscene, foul or boorish language that is inappropriate or improper for civilised discourse, directed or undirected.</li> <li>Target: Individual The target of a toxic comment is an individual</li> <li>Target: Group The target of a toxic comment</li> <li>Target: Other the target of an insult or an incite is not a person or a group of people is a group.</li> </ul>

ShareScore

40/100

Overall dataset sharing score

Score breakdown

These five areas show where the dataset supports — or may limit — practical reuse.

Stewardship
8
Harmonization
4
Access
20
Reuse readiness
8
Engagement
0