One of the biggest problems that advertisers face is finding the best social media influencers to promote their brands. In this industry, which is projected to hit 10 billion dollars by 2020, finding the right influencer is important. To tackle this problem, five UC Berkeley undergraduate students set out to help companies find influencers to help their brands by building PrISM (Predicting Influence in Social Media).
PrISM is a product that gives companies advanced analytics about influencers on Twitter. In addition to basic statistics such as number of favorites/retweets, PrISM provides a much more in depth analysis of an influencer’s profile to help brands reach their target audience. These advanced metrics include follower count over time and activity over time. This allows brands to see how an influencer grows and gives some perspective about how where the influencer is trending for the future. PrISM also uses different machine learning algorithms and techniques to get a follower demographic breakdown, run sentiment analysis, and perform topic modeling on a user’s tweets.
A demographic breakdown allows brands to see what ages and genders an influencer is reaching. Sentiment analysis enables brands to see how well an influencer reaches an audience. Topic modeling allows brands to see what topics an influencer likes to tweet about, so they can align themselves with appropriate influencers. After compiling these different analytics, we displayed these statistics and graphs in a beautiful, customizable card dashboard. Early reactions to PrISM show that it has a lot of potential and could prove very valuable to advertisers. We cannot wait to see where PrISM takes us next!
http://github.com/AdeelCheema/PrISM