Current publication sharing systems inherit from the Web 2.0 philosophy the idea that users can add reusable information to support other peers, enabling them to insert new resources and to tag the existing ones; but, in their current form, these systems suffer of some limitations, such as the lack of tools for supporting users during the creation and organization of their personal concept spaces, and the poor utilization of tags as information sources for producing personalized recommendations. In this paper we propose a model for organizing dynamic and customizable concept spaces, based on innovative structures, and we introduce a mechanism for recommendation, based on tags and mainly on the way in which users connect resources in their concept spaces. Adaptive recommendations are generated analyzing the users' concept spaces, and evaluating the similarities among them in order to reveal the similarity among their goals and perspectives.
