Project | Team | Results | Partners | Contact
Islamofobic Hate Detector (Ishade)
One of the most negative consequences of the expansion of social networks in recent years has been the exponential increase in hate speech, which is usually directed at already stigmatized audiences, based on ethnic or cultural differences, but also religious ones. Islam is one of the most affected by hate crimes in most European countries, including Spain, something that has been enhanced by the increase in arrivals of migrants, refugees and asylum seekers from the Middle East.
Hate speech has the potential to influence public opinion, thus being able to encourage discriminatory and intolerant attitudes, and even trigger violent episodes in the physical environment and all kinds of hate crimes against Muslims. In this context, it is understood that it is necessary to try to detect Islamophobic discourses that spread massively on the Internet, in order to understand and combat them.
The ISHADE project will develop and evaluate an Islamophobic hate speech detector propagated through social networks in the Spanish context.
To do this, natural language processing and machine learning techniques will be used, which will allow not only to automatically identify Islamophobic messages transmitted through social networks, but also to monitor them in real time and longitudinally, thus allowing statistics to be established and strategies to be generated. combat these messages, something that would allow the spiral of hate to be stopped and, at the same time, to counteract Islamophobic crimes.
OBJECTIVES
The general objective is to develop and evaluate a detector of Islamophobic hate speech in Spanish propagated in social networks, using natural language processing and machine learning techniques, and execute an awareness strategy using the knowledge generated.
Specific objectives:
• Create an ad-hoc database with examples of Islamophobic hate messages propagated on social networks, classified and manually validated, which will be used as a training corpus for the development of the prototype.
• Generate and evaluate the classifier prototype that will automatically detect Islamophobic hatred on a large scale in social networks in Spanish, making use of the ad-hoc generated corpus, as well as surface learning and deep learning techniques.
• Develop an awareness strategy based on the observations generated in the development of the database of the Islamophobic hate detector.
Con la colaboración de:
Project funded by: Fundación Pluralismo y Convivencia.