Last.fm, the radio to measure.

'Share taste instead of files' reads one of the payoffs of Last.fm , broadcaster streaming in London which carries updated techniques to customize his plays. Drawing on the experience of years developed spontaneously in mailing lists and forums, we have activated a collaborative filtering that attempts to perceive the likings and non-likings of users, simultaneously adjusting their playlists. In fact, the refinement of algorithms that combine the tastes of other similar products in the perennial temptation to induce the purchase has been characterized in recent years more than one online venture, not Amazon that for several years he has bought some successfully implemented , although there are also attempts to free as Audioscrobbler . In the case of Last.fm data is detected through the voluntary inclusion of a profile from the user, or simply based on the data of use, ie if a music is playing until the end has a high score, otherwise one low . These data over time nonetheless create a profile of preferences, compared with that of other users allows to predict with a good chance that other song will be welcome user. A broader perspective it is put in the hands of users the choices made so far by the industry. With 6,000 registered users and 30,000 songs in the database, Last.fm has had proof of the proper functioning of its scheme intuitive and easy to use, with several independent labels who have contacted seeing a promotional vehicle can be 'free' of their activities, impossible to carry on commercial radio.