> "... We were able to see the Neda video out of Iran trending well before CNN linked it in, and we’ve begun to refine our capabilities there to be able to pinpoint stories like that." He said part of this technique involves looking for links being shared by unlike people, because that means they have universal appeal.
This is why I fell in love with computer science. I can already imagine the use of a good shortest path algorithm combined with collaborative filtering (Netflix competition-type algorithms) being used to tackle this problem in real-time. You are given a unique url and 1000 twitter accounts that linked to it. You also have the entire twitter-follower digraph. Now design an algorithm to calculate the "universality" factor with the lowest Big O. Now do this 10k times a second for 10k unique urls.
This is why I fell in love with computer science. I can already imagine the use of a good shortest path algorithm combined with collaborative filtering (Netflix competition-type algorithms) being used to tackle this problem in real-time. You are given a unique url and 1000 twitter accounts that linked to it. You also have the entire twitter-follower digraph. Now design an algorithm to calculate the "universality" factor with the lowest Big O. Now do this 10k times a second for 10k unique urls.