大型图上的社区检测 [英] Community detection on a very large graph
问题描述
我有一个非常大的有向图(一个社交网络图),约有800万个节点.我想在同一社区上运行社区检测算法(它可以是重叠的或不重叠的).
我看过InfoMap,但是对于这样的图形来说,它太慢了-我可能要花几周的时间.BIGCLAM是斯坦福SNAP中的另一种实现,但仅适用于无向图.
我可以使用一台服务器,该服务器具有40个内核和128GB RAM(磁盘上的网络约为60GB).是否有任何实施或研究可以对我有所帮助?
Louvain方法很快!有一个实现: https://github.com/vtraag/louvain-igraph .>
您应该先安装图形库和numpy库.
I have a very large directed graph (a social network graph) with about 8 million nodes. I would like to run a community detection algorithm on the same (it can be overlapping or non-overlapping).
I have had a look at InfoMap but it is too slow for the size of such a graph - it might as well take weeks (i think). BIGCLAM is another implementation in Stanford SNAP but it is only for undirected graphs.
I have a server at my disposal with 40 cores and 128GB RAM (And my network on the disk is around 60GBs) which I can leverage. Does there exist any implementation or research that could help me?
Louvain Method is fast! There is an implementation: https://github.com/vtraag/louvain-igraph.
You are supposed to install graph library and numpy library first.
这篇关于大型图上的社区检测的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!