Several techniques for improving web search have been developed over the last years. Most existing approaches are still limited, mainly due to the absence of qualitative criteria for ranking results and insensitivity to user preferences for guiding the search. At the same time, defeasible argumentation evolved as a successful approach in AI to model commonsense qualitative reasoning with applications in many areas, such as agent theory, knowledge engineering and legal reasoning. This paper presents ARGUENET, a recommender system that classifies search results according to preference criteria declaratively specified by the user. The proposed approach integrates a traditional web search engine with a defeasible argumentation framework.