Pyspark Support

Not an issue, 

But any way this can be implemented/used in pyspark?

Would like to use this for cluster generation on Entity Resolution based deduplication pipeline. Currently using Connected Components Algorithm which is simpler, but very unstable with false positive edges.

Kind regards, 

Pablo