Ego Splitter Algorithm Implementation Problem

[PawnDragon]

Hi, through experiments(including test examples), I noticed that Ego Splitter algorithm in the overlapping community detection can not partition overlapping communities well. The reason is using connected components in the ego graph as the local clustering algorithm. As a result, when a ego graph of a node is a connected graph, it can not generate different personas (only one). I think the So I made some changes that applying louvain method to the local clustering and it worked. However, I can not ensure partition performance. If possible, we can conduct more tests and perfect this feature

[rjurney]

Thank you for this, we would appreciate a pull request!

[PawnDragon]

I have already submitted a pull request, please check it. Thanks a lot!