MIT researchers develop algorithm to predict trending topics on Twitter
The trending topics usually make their appearance on the basis of events occurring at a given time, or just boring based on users who want to popularize a term. And now, a couple of MIT researchers are ready to present a new algorithm that can predict says trending topic before it explodes.
The work is paid by the professor and his student Devravat Shah Stanislav Nikolov, which would anticipate the popularity of a topic with a half hour before and up to 95% accuracy, and might even get to “guess” something that will be of fashion until five hours later.
Of course, all this might be interesting, much to Twitter and its advertisers. According Ashish Goel, a member of the Technical Board of Twitter, “people go to Twitter to find out what is happening at the moment”, so this algorithm would help “accelerate some processes.” Among other things, Twitter would be interested to know what lies ahead in terms of fashion on their home page to sell advertising in between all the topics that will be as popular within the next few hours.
Shah’s work and his student Nikolov be presented in detail next week, amid the workshop Interdisciplinary Workshop on Information and Decision in Social Networks.algorithm, MIT, research, Trending Topic, Twitter