Web recommendation systems have become a popular means to improve
the usability of web sites. This paper describes the architecture of a rulebased
recommendation system and presents its evaluation on two real-life applications.
The architecture combines recommendations from different algorithms
in a recommendation database and applies feedback-based machine
learning to optimize the selection of the presented recommendations. The recommendations
database also stores ontology graphs, which are used to semantically
enrich the recommendations. We describe the general architecture of the
system and the test setting, illustrate the application of several optimization approaches
and present comparative results.