Information filtering using fuzzy models

Information filtering means delivery of various types of information items according to the needs of users. While information filtering extends information retrieval in the traditional sense, the former moreover includes selection of everyday information such as TV programmes, business news and electronic mail. This paper proposes the use of fuzzy technologies developed in information retrieval and classification, emphasizing how fuzzy features arise in different types of information items. In particular, the concepts of fuzzy prefiltering and postfiltering are discussed which include a variety of methods such as clustering, thesauri, user preferences, and so on. A view of a filtering system structure is shown which consists of filtering on the distributor side and that on the user side. Fuzzy and nonfuzzy methods are compared and advantages of fuzzy methods are shown.