Tel Aviv University researchers, led by Prof. Yuval Shavit, have developed a computer algorithm that has shown a 50% chance of identifying pop music stars based on file sharing networks like iMesh. The system, first developed by a grad student as his project for a master’s degree in computer science, analyzed 500 million requests by users for file downloads on the Gnutella file sharing network, and developed a list of “most likely to succeed” unsigned musicians and groups whose songs were being downloaded – without even listening to the songs. When the list was compared later to the U.S. Billboard popularity charts, as many as half the groups picked by the algorithm were found to have succeeded, either with a monster hit or a record contract.
For example, the system “discovered” Soulja Boy Tell’em and Sean Kingston weeks before their first hits even entered the Billboard Hot 100, and Shop Boyz before they were even signed onto a label. All three of these had grammy nominated/winning songs in 2007, and Shop Boyz Party Like a Rockstar was the number one downloaded ringtone in 2007. According to Prof. Shavit, there are many factors involved in success, including geography (where the artist is located, how often they play live, etc.) that the algorithm analyzes, and he believes that even greater accuracy will be possible with more work.
The grad student, Noam Kingstein, won a prize from TAU for the most commercially viable project, and at least one record company has already spoken to him. This is maybe the first time a system has been developed to allow record companies to profit from online music sharing, which has so far devastated sales.
e myself on my own blog, can’t I?
Entries (RSS)