Generally, an information filtering system can be on the server side or on the client side. The proposed fuzzy filtering approach can be used both ways. In client-sided filtering, the taxonomy may be in the form of user 's bookmarks, for example. The system creates the topic classifiers that the user uses to filter the results of a search engine, according to the method described. In serversided filtering, the taxonomy may be in the form of a web directory.
In order to provide an example application, we have developed a serversided filtering system on the Web using the Computers/Artificial Intelligence directory of the Open Directory Project (Dmoz). We created NB classifiers for 10 topics related to this directory, using about 40 web pages as training examples for each topic. The application is available from (ISLab). Through the interface, the user is able to create queries and rules in order to retrieve and filter web pages.