| Predicting Popularity of Online Content |
[Nov. 17th, 2008|02:21 pm] |
Urban legend turned science tells tales of a day when great machine secured away in the vault of every major record label will algorithmically interpret the waveform from a song and determine whether or not the song will be a hit. Spooky, no? But with that much in mind, it makes sense that efforts towards making similar predictions about the popularity of internet content wouldn't be too far behind.
Two researchers from HP's Social Computing lab, Bernardo Huberman and Gabor Szabo, just published a pretty in-depth study surrounding new content on YouTube and Digg, aimed at predicting the level of traffic and traction submissions might achieve. Unfortunately, a lot of this bounced up and over my head pretty quickly. However, there is some interesting analysis in here, including the optimal times of day to post new content and the expected length of time to measure general success by.
The pdf is here for general consumption. |
|
|
![[info]](http://l-stat.livejournal.com/img/userinfo.gif){kind=small}
rubicante_kid's journal