Label Propagation for Community Detection

see CUG’15 paper: Implementing a social-network analytics pipeline using
Spark on Urika XA

https://cug.org/proceedings/cug2015_proceedings/includes/files/pap158.pdf
https://cug.org/proceedings/cug2015_proceedings/includes/files/pap158-file2.pdf


Computation consists of these steps:
* Read in gzip files storing JSON-encoded tweets
* Find pairs of Twitter users that @mention each other
* Construct a graph from such users
* Run a label propagation algorithm on that graph
* Output the community structure resulting from label propagation
