This project is sparked by https://github.com/thangdnsf/BigCLAM-ApacheSpark
which also implements BigCLAM models proposed by Yang and Leskovec (2013),
I change most of the collectasmap and broadcast code into rdd join to make it more resource-efficent and robust.
I use this code to detect the communities of network which has tens of millions of nodes in my work,and it worked.
Important Notices of the code:
1.In Bigclam.scala,the graphpath file need to contain paires of edges in network whose lines are delimited by "\n",and whose node is delimited by "\t" like:
1\t2\n 3\t4\n
2.In Bigclam.scala,the nodeid need to be in range(0~max(num_nodes)-1),where num_nodes means the number of distinct nodes in the graph file
3.use sbt assembly to compile the program