Simplicial similarity and its application to hierarchical clustering