You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Leiden seems to be built like Louvain, in the sense that in the process of finding the best partition the nodes within a community are merged into a metanode, before performing again the nodes mouvment and etc.., here with the metanodes and so on until no improvement on the optimzed function. Like this we form a dendrogram giving a hierarchical representation of the graph with several levels of partitions.
Is there a way to recover the dendrogram from the Leiden aglo?
Thanks in advance,
Regards
Sami
The text was updated successfully, but these errors were encountered:
More detail can be found in their Algorithm 1, but here is the idea:
apply a graph-clustering algorithm to find the bottom clusters;
(a) compute the similarity between all pairs of bottom clusters;
(b) merge the two clusters that are the most similar, and update the similarities
between this new cluster and the existing ones;
(c) repeat step (2b) until all clusters have been merged into a single one.
The graph clustering algorithm in step 1 would of course be leiden algorithm
Helloc @vtraag,
Leiden seems to be built like Louvain, in the sense that in the process of finding the best partition the nodes within a community are merged into a metanode, before performing again the nodes mouvment and etc.., here with the metanodes and so on until no improvement on the optimzed function. Like this we form a dendrogram giving a hierarchical representation of the graph with several levels of partitions.
Is there a way to recover the dendrogram from the Leiden aglo?
Thanks in advance,
Regards
Sami
The text was updated successfully, but these errors were encountered: