Introducing groups to the MADCOW annotation system solves the privacy-collaboration problem for users. The proper matching between groups and users solved the group join problem for both groups’ owners and users. We used ontological and URL-based measures to execute the match. For ontological-based measures, MADCOW domains were used and linked with external knowledge repositories: ontologies. URL-based measure depends on calculating the number of URLs annotated both by group members and by MADCOW users external to the group as a method for quantifying shared interests. In this work, we describe the system, the problems, and their solutions, with reference to the design choices made.