This paper presents the classification process in a recommender system used for textual documents taken especially from web. The system uses in the classification process a combination of content filters, event filters and collaborative filters and it uses implicit and explicit feedback for evaluating documents.