Abstract
In this paper, we first discuss a methodology for the store and reuse of various types of information including design data. Architectures of such systems will consists of three components; (1) the extraction of information unites and the attachment of attributes; (2) data mining or information filtering; and (3) the customization of stored unites. In these components, we developed a prototype of an information filtering system on the basis of recommender techniques and a star-coordinates type of visualization technique. In this system, all data in database are mapped onto a 2D plane so that their coordinates represent the importance of the data. Control axes defined on the 2D plane can easily control the positions, that is, the importance of data. Our system enables users to recognize the overall tendency of dataset and to control recommendation results intuitively.
