johann-petrak
commited on
Commit
•
ce54256
1
Parent(s):
ac55e31
First updated version of datasheet
Browse files- datasheet.md +24 -24
datasheet.md
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
# GerMS-AT Dataset Datasheet
|
2 |
|
3 |
-
This file contains information about the GerMS-AT Dataset structured according to
|
4 |
["T. Gebru et.al. (2021): Datasheets for Datasets"](https://arxiv.org/abs/1803.09010)
|
5 |
|
6 |
If there is a need for additional information or clarification, please feel free to contact any
|
@@ -8,35 +8,35 @@ of the the maintainers of this repository.
|
|
8 |
|
9 |
### Motivation
|
10 |
|
11 |
-
* Purpose of dataset creation:
|
|
|
|
|
|
|
12 |
* Dataset creators:
|
|
|
|
|
13 |
* Funding of dataset creation:
|
|
|
14 |
|
15 |
### Composition
|
16 |
|
17 |
-
* Instance representation:
|
18 |
-
* Number of instances:
|
19 |
-
* Completeness/sampling:
|
20 |
* Data per instance:
|
|
|
|
|
|
|
21 |
* Label/target per instance:
|
22 |
-
*
|
23 |
-
|
|
|
24 |
* Recommended data splits:
|
25 |
-
*
|
26 |
-
*
|
27 |
-
* Presence of
|
28 |
-
|
29 |
-
* Identifyability of subpopulations:
|
30 |
-
* Identifyability of individuals:
|
31 |
-
* Presence of sensitive information:
|
32 |
-
|
33 |
-
### Collection Process
|
34 |
-
|
35 |
-
* Data associated with each instance:
|
36 |
-
* Data collection procedure:
|
37 |
-
* Sampling strategy:
|
38 |
-
* People involved in the data collection process:
|
39 |
-
* Collection period:
|
40 |
-
* Ethical review processes:
|
41 |
-
* Direct/indirect data collection:
|
42 |
|
|
|
1 |
# GerMS-AT Dataset Datasheet
|
2 |
|
3 |
+
This file contains information about the GerMS-AT Dataset partially structured according to
|
4 |
["T. Gebru et.al. (2021): Datasheets for Datasets"](https://arxiv.org/abs/1803.09010)
|
5 |
|
6 |
If there is a need for additional information or clarification, please feel free to contact any
|
|
|
8 |
|
9 |
### Motivation
|
10 |
|
11 |
+
* Purpose of dataset creation: the corpus was created in the context of [project FemDwell](https://www.ofai.at/projects/femdwell)
|
12 |
+
in order to build a machine-learning based assistant to help human content moderators
|
13 |
+
identify individual forum posts which may contain sexist or misogynist comments or whole fora, where an unusual large number of such posts
|
14 |
+
were created more easily.
|
15 |
* Dataset creators:
|
16 |
+
* [Austrian Research Institue for Artificial Intelligence](https://ww.ofai.at)
|
17 |
+
* [Standard Verlagsgesellschaft](https://about.derstandard.at/impressum/)
|
18 |
* Funding of dataset creation:
|
19 |
+
* The project was sponsored by [Vienna Business Agency](https://wirtschaftsagentur.at/)
|
20 |
|
21 |
### Composition
|
22 |
|
23 |
+
* Instance representation: JSON dictionary
|
24 |
+
* Number of instances: 7984
|
|
|
25 |
* Data per instance:
|
26 |
+
* `id`: a unique ID
|
27 |
+
* `text`: the text of the forum comment posted, with user names and real person names replaced by the string "{USER}" and web addresses, email address and
|
28 |
+
the like replaced by the string "{URL}"
|
29 |
* Label/target per instance:
|
30 |
+
* for each instance there is a dictionary which contains the label assigned by each of the annotators who were presented the comment. Annotators are
|
31 |
+
represented by anonymized annotator ids.
|
32 |
+
* the label assigned by each annotator is one of "0" (no sexism/misogyny present), "1" (mild), "2" (present), "3" (severe), "4" (extreme).
|
33 |
* Recommended data splits:
|
34 |
+
* the data is made available as a train and a test file, with the same split that was used in the [GermEval2024 GerMS-Detect](https://ofai.github.io/GermEval2024-GerMS/) shared task
|
35 |
+
* Presence of confidential information: there is no confidential information in the dataset
|
36 |
+
* Presence of offensive or otherwise problematic data: comments will contain sexist and misogynist remarks and may also contain other forms
|
37 |
+
of offensive or toxic remarks.
|
38 |
+
* Identifyability of subpopulations: comments have been made by readers of the online news web site
|
39 |
+
* Identifyability of individuals: all information that could identify a user or person has been removed or replaced with placeholders
|
40 |
+
* Presence of sensitive information: none
|
41 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
|