The research project NOHATE is set up for three years and divided into three phases: The first phase serves to identify and test a category system for the analysis of online comments and hate speech about refugees and migration, based on previous research as well as the experiences of the individual project partners from political and media backgrounds. During the second phase, this category system will be transfered into algorithms for the analysis of online comments in order to identify hate speech. The algorithm will be optimized and implemented in collaboration with all project partners. In the third phase, the algorithm will be implemented into a tool that translates the empirical classification into practical tools for the management of online communities and commentary sections.
Three partners work together in this research project: Freie Universität Berlin, Beuth University of Applied Sciences Berlin and VICO Research and Consulting.
Freie Universität Berlin is represented by Prof. Dr. Martin Emmer and Prof. Dr. Joachim Trebbe. Furthermore, Dr. Sünje Paasch-Colberg and Christian Strippel join the project as research assistants. The team combines research experience in the field of Internet usage and political online communication (Emmer and Strippel) with knowledge in the field of media, migration, and social integration (Trebbe and Paasch-Colberg). The team's main task includes the development of a theoretically well-founded category system for analyzing online comments and hate speech as well as for testing hypotheses about causes and dynamics of online debates.
Beuth University of Applied Sciences Berlin is represented by Prof. Dr. Alexander Löser and Prof. Dr. Felix Gers from the working group "Data Science". Their competences in explorative textual analysis (Löser) and knowledge presentation with neuronal networks, especially LSTMs (Gers), add research and developmental benefit to the project. The team investigates procedures for the analysis of hate speech by using their specifically developed data base INDREX as well as fundamental work on Long Short Term Neuronal Networks (LSTM). The team's main task within the project is to transfer the theoretically founded category system into a self-learning algorithm and to test and improve this algorithm.
VICO Research & Consulting was founded in Leinfeld-Echterdingen in 2005 and has become the leading German social media agency. Their social media monitoring tool "VICO Analytics" allows fast access to analyzing and monitoring online communication on social media platforms (Facebook, Twitter, forums, news sites, etc.) as well as an adaptation to individual needs. VICO guarantees an ideal coverage of the relevant data and combines the best possible methods and data providers. Due to the work of computer linguists and a semantic analysis of the collected data, the results are of unique precision.