Cluster analysis with deolistic graphs

Abstract

This work introduces a particular family of acyclic graphs referred to as Deolistic graphs. The formal definition of a Deolistic graph is presented together with a new clustering algorithm that uses logical sentences to reveal hidden structures in data. The sentences yielded by the algorithm can be generated by automatic means, considerably reducing the complexity of data analysis and time spent on the clustering process, not to mention the usefulness of its results. Furthermore, these sentences are rules that can be easily understood, interpreted and disclosed to all interested parties, thereby improving communication and reducing misunderstandings.