Datasets:

Modalities:
Tabular
Text
Formats:
csv
Languages:
German
ArXiv:
Libraries:
Datasets
pandas
License:
davanstrien HF staff commited on
Commit
58b4402
1 Parent(s): bf44216

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +42 -0
README.md ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ task_categories:
4
+ - text-classification
5
+ language:
6
+ - de
7
+ pretty_name: GAHD
8
+ ---
9
+
10
+ **NOTE** README copied from https://github.com/jagol/gahd
11
+
12
+ This repository contains the dataset from our NAACL 2024 paper "Improving Adversarial Data Collection by Supporting Annotators: Lessons from GAHD, a German Hate Speech Dataset".
13
+
14
+ `gahd.csv` contains the following columns:
15
+ - `gahd_id`: unique identifier of the entry
16
+ - `text`: text of the entry
17
+ - `label`: `0` = "not-hate speech", `1` = "hate speech"
18
+ - `round`: round in which the entry was created
19
+ - `split`: "train", "dev", or "test"
20
+ - `contrastive_gahd_id`: `gahd_id` of its contrastive example
21
+
22
+ `gahd_disaggregated.csv` contains the following additional columns:
23
+ - `source`:
24
+ - if annotators entered the entry via the Dynabench interface: `dynabench`
25
+ - if the entry was translated from the Vidgen et al. 2021 dataset: `translation`
26
+ - if the entry stems from the Leipzit news corpus: `news`
27
+ - `model_prediction`: label predicted by the target model, `0` or `1`
28
+ - `annotator_id`: unique identifier of the annotator that created the entry
29
+ - `annotator_labels`: a string containing a forward slash-separated list of all labels by annotators
30
+ - `expert_labels`: `0` or `1` if an expert annotator annotated the entry, otherwise empty
31
+
32
+ When using GAHD, please cite our preprint on Arxiv:
33
+ ```
34
+ @misc{goldzycher2024improving,
35
+ title={Improving Adversarial Data Collection by Supporting Annotators: Lessons from GAHD, a German Hate Speech Dataset},
36
+ author={Janis Goldzycher and Paul Röttger and Gerold Schneider},
37
+ year={2024},
38
+ eprint={2403.19559},
39
+ archivePrefix={arXiv},
40
+ primaryClass={cs.CL}
41
+ }
42
+ ```