Symbolic Data Clustering

Abstract

In data mining, we generate class/cluster models from large datasets. Symbolic Data Analysis (SDA) is a powerful tool that permits dealing with complex data (Diday, 1988) where a combination of variables and logical and hierarchical relationships among them are used. Such a view permits us to deal with data at a conceptual level, and as a consequence, SDA is ideally suited for data mining. Symbolic data have their own internal structure that necessitates the need for new techniques that generally differ from the ones used on conventional data (Billard & Diday, 2003). Clustering generates abstractions that can be used in a variety of decision-making applications (Jain, Murty, & Flynn, 1999). In this article, we deal with the application of clustering to SDA.