MINING SOCIAL MEDIA FOR CONFLICT PREVENTION AND RESOLUTION

Abstract

The power of social media such as Twitter, Facebook, Instagram, LinkedIn, etc. in our daily lives cannot be underestimated. Governments have been toppled and countries destabilized as a result of sentiments expressed by citizens on social media. In this paper, we show that mining Twitter Follower/Friend network structure and data can be a powerful method to recognize the needs, sentiments, opinions and interests of the citizenry. Hierarchical clustering and Partition around Medoids were used. It was discovered that the Twitter community in Ghana takes delight in discussing political parties and personalities instead of pressing issues like corruption and unemployment. Follower/Friends network analysis was used to discover influential “e-people” who could serve as potential mediators during conflict situations. This method is aimed at identifying the most influential people in the Ghanaian Twitter Community and to discover what most people are complaining about through their tweets. This can be used to avoid a replication of the “Arab Spring” elsewhere. Possible Mediators can also be discovered. We propose an inexpensive but effective method to help prevent and resolve the rampant conflicts in the World that arise due to neglect of citizens by their governments. Advertisers, policy makers and political parties also stand to benefit from this approach.