OpenBib is a next generation discovery infrastructure for libraries. On the frontend side it features RSS feeds, mashups, tagging, tag clouds, use analyses, drilldowns as well as catalogue enrichment and much more. The entire infrastructure relies on the central priciples of REST and the Semantic Web.
A lot of different data sources - ranging from library catalogues to collections to OAI repositories - can be imported in its local data store (PostgreSQL database + Xapian search engine). Any combination of these data sources can be exposed as a separate web portal with minimum work using OO inheritance of templates on different abstraction levels.
As a use case see the union catalogue of the University of Cologne, Germany: