Open source intelligence gathering for hate speech in Kenya
Adhi, Banchale Gufu
MetadataShow full item record
The Internet has been celebrated for its ability to erode barriers between nations. Social media is a powerful medium that can unite, inform, and move people. One post can start a chain of events that changes the world. It gives users fast access to and sharing of information and facilitates ease of communication. However, the Internet allows for a lot of negativity as well. There has been an increase in hate speech activities on social media in the Kenyan cyber space. The National Cohesion and Integration Commission (NCIC) was established to facilitate and promote equality of opportunity, good relations, harmony and peaceful co-existence between persons of the different ethnic and racial communities of Kenya, and to advise the Government on all aspects thereof (Act No, 12, 2008). In particular, the NCIC Act of 2008 is mandated to curb hate speech. This research studied existing hate speech detection tools in use by NCIC, then identified gaps and challenges faced. A technical solution (tool for analyzing hate speech) was proposed that can be implemented by the NCIC and the government to respond to hate-speech cases perpetrated through social media platforms. The developed tool tracked challenges and gaps in the existing tools currently in use by NCIC for hate speech monitoring, detection and analysis. Due to the differences in Application Programming Interface (API) implementation on the variety of social media platforms used in Kenya, the scope of this research is limited to Twitter. This research employed the use of predictive analytics for text classification using Naïve Bayes. A tool that uses the predictive model in assistance to detection of hate-speech online was developed to conceptualize the solutions discussed in this research.