Skip to content

danielgbaena/spanish-hope-multidomain

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Annotation guidelines

The SpanishHopeMultidomain dataset consist of 2,000 tweets annotated as hs (hope speech) or nhs (non-hope speech). A text was considered hs or nhs if it matched the requirements from the annotation guidelines listed below.

Index

Annotation instructions

General instructions

Each text corresponds to one tweet and every tweet would be tagged only if it makes complete sense by itself and, therefore, if it is not necessary to read any additional tweets from the comments, retweets or review the rest of tweets in the thread, in case our tweet could be part of a Twitter thread of tweets. Nevertheless, if in order to understand a tweet it is necessary to know some popular news, such as the murder of Samuel Luiz in La Coruña (Spain) or the well-known Spanish “ley trans” draft law presentation, to name a few examples, we will consider our tweet to be comprehensive by itself and then we will proceed to tag it within a category other than None. Finally, in all cases we will only tag each tweet in a single category.

Working with the spreadsheet

The text column will show the content of each tweet, the category column will be filled in with the category that the annotator will associate with the tweet and the topic column will be associated with the theme of the text. Thus, all tweets will be tagged using the categories described in the next section.

Categories

All possible categories are listed below:

  1. Hope speech (hs).
  2. Not hope speech (nhs).

For each category, a definition is provided reflecting when a tweet should be considered as belonging to it. In addition, several examples are shown to make them more understandable and the choice of tags for all examples is explained.

Hope speech

A text was annotated as hs (hope speech) if:

  1. It explicitly supports the social integration of minorities.
  2. It is a positive inspiration for a vulnerable community.
  3. It explicitly encourages people who might find themselves in a situation.
  4. It unconditionally promotes tolerance. Below, you will find several examples, accompanied by their explanation of label choice.

Case 1

It explicitly supports the social integration of minorities.

Example

He leído a gente diciendo que a las lesbianas le pueden gustar los hombres porque hay mujeres trans?? Si eres una mujer trans eres una MUJER, no eres ni un hombre ni una mujer a medias ni leches. . Y lo mismo, si a un chico trans le gusta una mujer es hetero, porque es UN CHICO.

Case 2

It is a positive inspiration for a vulnerable community.

Example

¿Qué necesidad tiene la gente de decirme que estoy gorda? 🙄. Bro, no me importa tu comentario. Estoy bien y me siento bien con mi cuerpo actualmente.

Case 3

It explicitly encourages people who might find themselves in a situation.

Example

Eee? Gorda?? Vos? Vos estás hermosa, ayer, hoy y siempre. No le hagas caso a esos porque se ve que necesitan más aumento o no sé, capaz que los lentes son de la feria

Case 4

It unconditionally promotes tolerance.

Example

Sí, todo mi apoyo. Pero sí, debe de haber equilibrios. Yo, como persona🏳️‍🌈 GAY, debo tener los mismos derechos que los heteros. El equilibrio es democrático. No caigamos en clasismos, favoritismos y prejuicios. Son derechos fundamentales. El INE, no puede, ni debe disponer derecho

Not hope speech

On the other hand, a text will be annotated as nhs (not hope speech) if:

  1. It does does not express any positive sentiment.
  2. It explicitly seeks violence.
  3. It uses insults.

Case 1

It does does not express any positive sentiment.

Example

Un inmigrante en paro devuelve 400 euros que se encontró en una caja olvidada entre dos coches de la calle Pérez Galdós de Logroño

Case 2

It explicitly seeks violence.

Example

Tío cuando van a dejar de quejarse de la "inclusión forzada"?. Es negra? inclusión forzada. Es homosexual? inclusión forzada. Es mujer? inclusión forzadísima. Hermano callaos ya la puta boca os falta decir que la sirenita fue inclusión forzada de las pelirrojas en su momento

Case 3

It uses insults.

Example

No tengo ni puta idea son unos maricones hijos de la gran puta los que fushean

Topics

All possible topics are listed below:

  1. LGBT (lgbt).
  2. Obesity (obesity).
  3. Racism (racism).

For each topic, a definition is provided reflecting when some text should be considered as being part of it. In addition, several examples are shown to make them more understandable and the choice of tags for all examples is explained.

LGBT example

He leído a gente diciendo que a las lesbianas le pueden gustar los hombres porque hay mujeres trans?? Si eres una mujer trans eres una MUJER, no eres ni un hombre ni una mujer a medias ni leches. . Y lo mismo, si a un chico trans le gusta una mujer es hetero, porque es UN CHICO.

Obesity example

Eee? Gorda?? Vos? Vos estás hermosa, ayer, hoy y siempre. No le hagas caso a esos porque se ve que necesitan más aumento o no sé, capaz que los lentes son de la feria

Racism example

Un inmigrante en paro devuelve 400 euros que se encontró en una caja olvidada entre dos coches de la calle Pérez Galdós de Logroño

About

Automatic hope speech detection Spanish multidomain dataset

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published