As social media becomes increasingly important as a data source for the purposes of machine learning, finding a brand new method for analyzing the Twitter microblogging platform is very compelling. Tauid Zaman, assistant professor at MIT’s Sloan School of Management, developed a probabilistic model for the spread of an individual tweet in the twitterverse. By observing the timing of the retweets of a tweet, he is able to predict the total number of retweets the tweet will receive. This work is done in collaboration with Emily Fox and Eric Bradlow.
The algorithm takes advantage of the fact that you can model a systematic, repeatable behavior centered around the process of retweeting. You can experiment with the algorithm yourself over at the Twoija: Retweet Oracle website. The site charts (see sample above) the retweets of about 50 sample tweets from politicians and celebrities in order to show how they match up with prediction.
There is a practical side of all this. The research could help advertisers better understand the spread of ideas and give users a basis to monetize their tweets. Zaman’s other research focus using social media determines someone’s influence on Twitter. He found that during major events , one Twitter user becomes a “superstar.” Because of the high activity of their followers, these superstars garner substantially more retweets.