Distributed, real-time computation of community preferences

Abstract

We describe the integration of smart digital objects with Hebbian learning to create a distributed, real-time, scalable approach to adapting to a community's preferences. We designed an experiment using popular music as the subject matter. Each digital object corresponded to a music album and contained links to other music albums. By dynamically generating links among digital objects according to user traversal patterns, then hierarchically organizing these links according to shared metadata values, we created a network of digital objects that self-organized in real-time according to the preferences of the user community. Furthermore, the similarity between user preferences and generated link structure was more pronounced between collections of objects aggregated by shared metadata values.