Is clustering a special case of similarity join?

Question

Can clustering be treated as a special case of similarity join?

Clustering [...] organizes data instances into similarity groups, called clusters such that the data instances in the same cluster are similar to each other and data instances in different clusters are very different from each other.

This quote from Bing Liu, Web Data Minig, Exploring Hyperlinks, Contents, and Usage Data, Second Edition describes clustering as a way of grouping similar objects.

Also, join-around sounds very similar to clustering to me since it returns all the objects within a specified distance from another object.

Answer 1

For some algorithms, such an operation can be useful.

In particular, DBSCAN is built exactly in this kind of operation. However:

1. I am not aware of any database that would efficiently implement this. Specialized tools like ELKI drastically outperform databases here.
2. Computing the join may need up to O(n²) memory. These memory requirements may kill you. DBSCAN can be implemented with just O(n) memory.