This paper presents an Evolutionary Algorithm used to search for similarities in a music score represented as a graph. We
show how the graph can be searched for similarities of different kinds using interchangeable similarity measures based on
viewpoints. A segmentation algorithm using the EA for automatically finding structures in a score based on a specific-to-general ordering
of the viewpoints is proposed. As an example a fugue by J. S. Bach is analysed, revealing its extensive use of inner resemblance.