Dynamic Linking of Smart Digital Objects Based on User Navigation Patterns

We discuss a methodology to dynamically generate links among digital objects by means of an unsupervised learning mechanism which analyzes user link traversal patterns. We performed an experiment with a test bed of 150 complex data objects, referred to as buckets. Each bucket manages its own content, provides methods to interact with users and individually maintains a set of links to other buckets. We demonstrate that buckets were capable of dynamically adjusting their links to other buckets according to user link selections, thereby generating a meaningful network of bucket relations. Our results indicate such adaptive networks of linked buckets approximate the collective link preferences of a community of users.