Counter-bot: Artificial Intelligence combating racist hate speech online

The aim of the project is to create a scientifically sound basis for the implementation of a counter-bot that automatically generates effective counter-speech against racist hate speech online. The inductively identified indications of characteristic structures of hate speech online will be analyzed linguistically and statistically to provide a useful starting point for the future implementation of a counter-bot.

For this purpose, ZARA will provide guidance for drawing a sample of suitable postings from the large number of racist postings on Twitter already collected by ZARA’s counselling center #GegenHassimNetz. During the first step of the analysis a category scheme will be developed to make the relevant characteristics of the postings accessible for statistical analysis. In a second step, characteristic patterns in the sequence of postings will be scientifically identified on the basis of the categories previously determined. These two sub-steps allow for an analytical understanding of the structure of hate postings and the effect of counter-speech. Such patterns will play a pivotal role in determining the mode of operation of the counter-bot and will become an integral part of the functional specification document, which is the basis for the technical realization of the counter-bot in a follow-up project.

The final specification document (German only) is available in the download area below.