Efficient Clustering of High-Dimensional Data Sets with Application to Reference Matching 

- Canopies can be applied to many domains and used with a variety of clustering approaches, including 

            Greedy Agglomerative Clustering,

            K-means and 

            Expectation-Maximization.

- we do not calculate the distance between two points that never appear in the same canopy, i.e. we assume their distance to be infinite.