Abstract
In the emerging markets for digital content distribution, traditional models for content programming still predominate. But there is a large potential for innovation and improvement in content selection to better meet consumers' demand for entertainment and news. We have developed a new paradigm and supporting technology to customize audience content to their specific preferences. One can identify several selection models for groups of individuals with similar interests in music, video, or other multimedia content to jointly customize a distribution channel. Our approach represents a balance between the two most widespread models available today, namely broadcasting and individual playback such as CD/DVD players. Using technologies such as data mining, multicasting and smart players, our model gives listeners access to automatic shared playlists. This kind of customized narrowcasting is especially applicable to distribution of content for which there is high demand for repeat listening or viewing, while at the same time being very idiosyncratic. Our approach can provide significant advantages to consumers, distribution channels, content owners and advertisers alike. We present the basic collaborative content programming algorithms and describe initial experiences with this new paradigm. Specifically, we describe KARC, a prototype of an Internet multichannel virtual radio station environment deployed at the IBM Almaden Research Center.