Around the middle of March we started analyzing our traffic with regards to the NCAA 2013 championship. To do that, we chose the top thirteen teams at the time (Gonzaga, Michigan State, Indiana, Michigan, Georgetown, Nashville, Kansas, Louisville, Duke, Miami, St. Mary’s, Kennesaw State and La Salle) and built custom categories for identifying them. DG-Peer39’s system currently crunches about fifty billion requests per day. Each request represents a web page which is about to be delivered to an internet user, which is sent to DG-Peer39 for analysis.
We enabled the system to identify web pages referencing each of the teams specifically in the context of basketball. These pages were automatically analyzed on a deep semantic level across three dimensions: safety, quality and topic. These massive amounts of analyzed traffic also provide us with a unique opportunity to glean insights and intelligence on current internet trends at large.
The graph below shows the request volume we received for each of the teams, as a fraction of the total number of March-Madness requests. The percentages are the relative share of each team in the total requests.
To further analyze this data, we sampled several tens of thousands of random webpages referencing Michigan and Louisville, the... Read more