On May 21, we were joined by Chris Orwa from iHub Research, based in Nairobi. The Center has worked with the team at iHub Research before, and we were excited to feature them as our first Innovation & Learning Webinar guest from the “field”!
Chris gave a fascinating presentation about the Umati project, which was implemented in 2013 to understand the use of dangerous speech in the Kenyan online space. Based on Susan Benesch’s framework for identifying dangerous speech, the iHub team developed a methodology for monitoring content in 8 local languages on blogs, forums, Facebook, Twitter, and online newspapers. From October 2012-November 2013, 7,000+ incidents of hate speech were captured. This evidence was then shared with organizations in Kenya that were responsible for countering hate speech.
Based on the information they gathered, the iHub team was able to analyze behavior differences on different platforms (i.e. Facebook versus Twitter), as well as form some initial hypotheses about the relationship between local language use and the degree of hate speech.
In Phase II of Umati, the team is working on automating data collection (rather than relying on a team of human monitors), building classifiers for automatic capture of dangerous speech, and implementing the project in Nigeria, which offers a similar local context, for their 2015 elections.
Check out our Hangout with Chris here!