DATA4CHANGE helps Civil Society Organisations create powerful data-driven advocacy projects by connecting them with global and local experts in data, design, tech and journalism to forge real change.
For the 2019 workshop in Nairobi, my team was tasked with creating a platform that could help with the identification and mitigation of hate speech propagation, specific to South Sudan.
Our systems design thinking and brainstorming resulted in a multimodal system of interventions (SIFTA), which can work on both a local and global scale. In prototyping one of these modules, I used VADER sentiment analysis alongside ACLED conflict data and FB Graph API data to model the toxicity of comment threads accompanying scraped articles from the Sudan Tribune. The aim of the tool is to understand the loudest voices—people who try to derail conversations using hateful terms. Moreover it aims to identify lexical trends and add them to a live database of hateful words and recurring media assets that frequently show up alongside hate speech/disinformation.