ABSTRACT
In this paper we propose a new method for the clustering of chemical databases based on the representation of the measures of structural similarity on multidimensional spaces. The proposed method allows the tuning of the clustering process by means of the selection of the number of spaces, the normal vectors and the sensibility of the projection process. The structural similarity of each element with regard to the elements of the database is projected on to the defined spaces generating clusters that represent the pattern and diversity of the database and whose size and characteristics can be easily adjusted.
