Datasets:

Modalities:
Text
Formats:
csv
ArXiv:
Libraries:
Datasets
Dask
License:
nedjmaou commited on
Commit
3c6f007
1 Parent(s): d2c919b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +40 -0
README.md CHANGED
@@ -1,3 +1,43 @@
1
  ---
2
  license: mit
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
  ---
4
+ # Dataset of our EMNLP 2019 Paper (Multilingual and Multi-Aspect Hate Speech Analysis)
5
+ For more details about our dataset, please check our paper:
6
+
7
+ @inproceedings{ousidhoum-etal-multilingual-hate-speech-2019,
8
+ title = "Multilingual and Multi-Aspect Hate Speech Analysis",
9
+ author = "Ousidhoum, Nedjma
10
+ and Lin, Zizheng
11
+ and Zhang, Hongming
12
+ and Song, Yangqiu
13
+ and Yeung, Dit-Yan",
14
+ booktitle = "Proceedings of EMNLP",
15
+ year = "2019",
16
+ publisher = "Association for Computational Linguistics",
17
+ }
18
+
19
+ (You can preview our paper on https://arxiv.org/pdf/1908.11049.pdf)
20
+
21
+ ## Clarification
22
+ The multi-labelled tasks are *the hostility type of the tweet* and the *annotator's sentiment*. (We kept labels on which at least two annotators agreed.)
23
+
24
+ ## Taxonomy
25
+ In further experiments that involved binary classification tasks of the hostility/hate/abuse type, we considered single-labelled *normal* instances to be *non-hate/non-toxic* and all the other instances to be *toxic*.
26
+
27
+ ## Dataset
28
+ Our dataset is composed of three csv files sorted by language. They contain the tweets and the annotations described in our paper:
29
+
30
+ the hostility type *(column: tweet sentiment)*
31
+
32
+ hostility directness *(column: directness)*
33
+
34
+ target attribute *(column: target)*
35
+
36
+ target group *(column: group)*
37
+
38
+ annotator's sentiment *(column: annotator sentiment)*.
39
+
40
+ ## Experiments
41
+
42
+ To replicate our experiments, please see https://github.com/HKUST-KnowComp/MLMA_hate_speech/blob/master/README.md
43
+