Datasets:
Tasks:
Text Classification
Formats:
csv
Languages:
Portuguese
Size:
10K - 100K
ArXiv:
Tags:
hate-speech-detection
License:
victoriadreis
commited on
Commit
•
35d58b6
1
Parent(s):
23e9328
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,136 @@
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
3 |
---
|
4 |
+
---
|
5 |
+
license: cc-by-4.0
|
6 |
+
annotations_creators:
|
7 |
+
- crowdsourced
|
8 |
+
language_creators:
|
9 |
+
- crowdsourced
|
10 |
+
language:
|
11 |
+
- pt
|
12 |
+
multilinguality:
|
13 |
+
- monolingual
|
14 |
+
size_categories:
|
15 |
+
- 10K<n<100K
|
16 |
+
source_datasets:
|
17 |
+
- original
|
18 |
+
task_categories:
|
19 |
+
- text-classification
|
20 |
+
task_ids: []
|
21 |
+
pretty_name: TuPy-Dataset
|
22 |
+
language_bcp47:
|
23 |
+
- pt-BR
|
24 |
+
tags:
|
25 |
+
- hate-speech-detection
|
26 |
+
configs:
|
27 |
+
- config_name: multilabel
|
28 |
+
data_files:
|
29 |
+
- split: train
|
30 |
+
path: multilabel/multilabel_train.csv
|
31 |
+
- split: test
|
32 |
+
path: multilabel/multilabel_test.csv
|
33 |
+
- config_name: binary
|
34 |
+
data_files:
|
35 |
+
- split: train
|
36 |
+
path: binary/binary_train.csv
|
37 |
+
- split: test
|
38 |
+
path: binary/binary_test.csv
|
39 |
+
---
|
40 |
+
|
41 |
+
# Portuguese Hate Speech Dataset (TuPy)
|
42 |
+
|
43 |
+
The Portuguese hate speech dataset (TuPy) is an annotated corpus designed to facilitate the development of advanced hate speech detection models using machine learning (ML)
|
44 |
+
and natural language processing (NLP) techniques. TuPy is comprised of 10,000 (ten thousand) unpublished, annotated, and anonymized documents collected
|
45 |
+
on Twitter (currently known as X) in 2023.
|
46 |
+
This repository is organized as follows:
|
47 |
+
|
48 |
+
```sh
|
49 |
+
root.
|
50 |
+
├── binary : binary dataset (including training and testing split)
|
51 |
+
├── multilabel : multilabel dataset (including training and testing split)
|
52 |
+
└── README.md : documentation and card metadata
|
53 |
+
```
|
54 |
+
## Security measures
|
55 |
+
To safeguard user identity and uphold the integrity of this dataset, all user mentions have been anonymized as "@user," and any references to external websites have been omitted
|
56 |
+
|
57 |
+
## Annotation and voting process
|
58 |
+
To generate the binary matrices, we utilized a simple voting process. Each document underwent three separate evaluations.
|
59 |
+
If a document received two or more identical classifications, the assigned value was set to 1; otherwise, it was marked as 0.
|
60 |
+
The annotated raw data can be accessed in the [project repository](https://github.com/Silly-Machine/TuPy-Dataset).
|
61 |
+
The following table offers a brief summary of the annotators' profiles and qualifications:
|
62 |
+
|
63 |
+
#### Table 1 – Annotators
|
64 |
+
|
65 |
+
| Annotator | Gender | Education | Political | Color |
|
66 |
+
|--------------|--------|-----------------------------------------------|------------|--------|
|
67 |
+
| Annotator 1 | Female | Ph.D. Candidate in civil engineering | Far-left | White |
|
68 |
+
| Annotator 2 | Male | Master's candidate in human rights | Far-left | Black |
|
69 |
+
| Annotator 3 | Female | Master's degree in behavioral psychology | Liberal | White |
|
70 |
+
| Annotator 4 | Male | Master's degree in behavioral psychology | Right-wing | Black |
|
71 |
+
| Annotator 5 | Female | Ph.D. Candidate in behavioral psychology | Liberal | Black |
|
72 |
+
| Annotator 6 | Male | Ph.D. Candidate in linguistics | Far-left | White |
|
73 |
+
| Annotator 7 | Female | Ph.D. Candidate in civil engineering | Liberal | White |
|
74 |
+
| Annotator 8 | Male | Ph.D. Candidate in civil engineering | Liberal | Black |
|
75 |
+
| Annotator 9 | Male | Master's degree in behavioral psychology | Far-left | White |
|
76 |
+
|
77 |
+
## Data structure
|
78 |
+
A data point comprises the tweet text (a string) along with thirteen categories, each category is assigned a value of 0 when there is an absence of aggressive or hateful content and a value of 1 when such content is present. These values represent the consensus of annotators regarding the presence of aggressive, hate, ageism, aporophobia, body shame, capacitism, lgbtphobia, political, racism, religious intolerance, misogyny, xenophobia, and others. An illustration from the multilabel TuPy dataset is depicted below:
|
79 |
+
|
80 |
+
```python
|
81 |
+
{
|
82 |
+
text: "e tem pobre de direita imbecil que ainda defendia a manutenção da política de preços atrelada ao dólar link",
|
83 |
+
aggressive: 1, hate: 1, ageism: 0, aporophobia: 1, body shame: 0, capacitism: 0, lgbtphobia: 0, political: 1, racism : 0,
|
84 |
+
religious intolerance : 0, misogyny : 0, xenophobia : 0, other : 0
|
85 |
+
}
|
86 |
+
```
|
87 |
+
|
88 |
+
# Dataset content
|
89 |
+
|
90 |
+
Table 2 provides a detailed breakdown of the dataset, delineating the volume of data based on the occurrence of aggressive speech and the manifestation of hate speech within the documents
|
91 |
+
|
92 |
+
#### Table 2 - Count of non-aggressive and aggressive documents
|
93 |
+
|
94 |
+
| Label | Count |
|
95 |
+
|----------------------|--------|
|
96 |
+
| Non-aggressive | 8013 |
|
97 |
+
| Aggressive - Not hate| 689 |
|
98 |
+
| Aggressive - Hate | 1298 |
|
99 |
+
| Total | 10000 |
|
100 |
+
|
101 |
+
Table 3 provides a detailed analysis of the dataset, delineating the data volume in relation to the occurrence of distinct categories of hate speech.
|
102 |
+
|
103 |
+
#### Table 3 - Hate categories count
|
104 |
+
|
105 |
+
| Label | Count |
|
106 |
+
|--------------------------|-------|
|
107 |
+
| Ageism | 53 |
|
108 |
+
| Aporophobia | 61 |
|
109 |
+
| Body shame | 120 |
|
110 |
+
| Capacitism | 92 |
|
111 |
+
| LGBTphobia | 96 |
|
112 |
+
| Political | 532 |
|
113 |
+
| Racism | 38 |
|
114 |
+
| Religious intolerance | 28 |
|
115 |
+
| Misogyny | 207 |
|
116 |
+
| Xenophobia | 70 |
|
117 |
+
| Other | 1 |
|
118 |
+
| Total | 1298 |
|
119 |
+
|
120 |
+
# BibTeX citation
|
121 |
+
|
122 |
+
This dataset can be cited as follows:
|
123 |
+
|
124 |
+
```pyyhon
|
125 |
+
@misc {silly-machine_2023,
|
126 |
+
author = { {Silly-Machine} },
|
127 |
+
title = { TuPy-Dataset (Revision de6b18c) },
|
128 |
+
year = 2023,
|
129 |
+
url = { https://huggingface.co/datasets/Silly-Machine/TuPy-Dataset },
|
130 |
+
doi = { 10.57967/hf/1529 },
|
131 |
+
publisher = { Hugging Face }
|
132 |
+
}
|
133 |
+
```
|
134 |
+
|
135 |
+
# Acknowledge
|
136 |
+
The TuPy project is the result of the development of Felipe Oliveira's thesis and the work of several collaborators. This project is financed by the Federal University of Rio de Janeiro ([UFRJ](https://ufrj.br/)) and the Alberto Luiz Coimbra Institute for Postgraduate Studies and Research in Engineering ([COPPE](https://coppe.ufrj.br/)).
|