Datasets:

Modalities:
Tabular
Text
Formats:
json
Libraries:
Datasets
pandas
License:
File size: 654 Bytes
2fb416e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
---
license: cc-by-sa-3.0
---

# ToxicSK

Author: Zuzana Sokolová


Corpus of toxic speech in social networks


The dataset contains 8 840 manually annotated comments from Facebook. Specifically, it includes:

- 4 420 comments labeled as "toxic" (value 1)
- 4 420 comments labeled as "non-toxic" (value 0)

The comments were annotated by a group of PhD students and assistant professors from the Technical University of Košice. The annotators' native language is Slovak. Each comment was annotated only once, and the labeling is based on the personal opinion of the annotator.

- Total number of sentences: 8 697
- Average sentence length: 9.40 words