We live in a world where virtually any song can be acquired at the click of a button. Some of us resort to file sharing, while others pay for subscriptions to MP3 download sites. Whichever your course, you're still constrained by the untrustworthy search function: If you can't remember the name, often you can't find the song. So how do you discover new music? You can't rely on the radio (who really has the time any more?), you can't read music reviews, and you certainly can't trust your whimsical friends. Not to worry- now we have new intelligent music technologies to help you through this misery!
One of the more popular methods, introduced by retail sites like Amazon.com, was collaborative filtering. The site tracks the shopping habits of customers and uses the collected data to recommend products to customers with similar shopping patterns. For example, if you purchased X and Y, Amazon would notice that the last 500 customers to buy X and Y also bought Z and would recommend you buy Z too.
This is an impressive tool, but it is one that demands from the outset a wide and diverse audience from which to track shopping patterns. It also relies on the idea that if you like Martha Stewart and Stephen King, you'll invariably like Harry Potter too. We're all much too fickle for that to apply to us!
However, there are some digital music sites that apply the collaborative filtering tool. MusicMatch Jukebox's newly released radio stations are just one example: They stream songs just for you, based on your (and the rest of their user base's) listening habits. Other sites, though, have introduced their own unique and fascinating technologies to recommend music that you're sure to love.
So what does go on behind the scenes? How do they know what music you'll like? MuBu, moodLogic, Listen, and Gigabeat are four of the leaders in this field, and they recently let us in behind closed doors to reveal some of their music recommendation secrets.
