Mapping Influence on Social Networks

Afghanistan, Libya, Burkina Faso, Mali, Senegal


Context and issues

Understanding interactions between main influencers on social media along with their opinions about the topics they discuss is key for public decision-takers, in order to know what kind of messages populations are exposed to in a given country.


Solution

Out of a seed list of well-known influencers, Masae runs a “snowball” algorithm browsing their network to identify other influencers connected with them. An intense study of their connections (followers/followings), their mentions and their retweets is conducted (including the topics mentioned) to build an influence network.

This network allows to map the country-wide media landscape and detect communities with strong intra-connections.

More advanced analysis include state-of-the-art NLP outputs, such as topic modelling, sentiment or hate speech detection with machine learning algorithms developed by Masae in many different languages (English, French, Amharic, Somali, local dialects of Arabic).





Masae provides an online platform to visualise the interaction network among influencers and main topics discussed.