Gephi forums

Work is currently done to bring minor and major improvements to the statistics. For instance Connected Components and Improve GraphDistance statistics are already confirmed blueprints, thanks to Patrick.

Would be nice if we could list new metrics and algorithms that could be implemented and evaluate the priorities.

I propose
- Eigenvector centrality
- Hierarchical clustering or KMeans, with various distance functions. See wikipedia

Share your ideas now

I thought computing all articulation points (nodes whose removal increases the number of weakly connected components) could be interesting, or bridges (same concept but over edges).

http://en.wikipedia.org/wiki/Articulation_Points

Anyone have experience with hierarchical clustering algorithms. A few algorithms I've found will a create a complete top-to-bottom hierarchy, such that the top item is the set of all nodes and the leaves are singleton nodes. Is that something users would be interested?

Or would it be more useful if the top of the hierarchy were well defined clusters of nodes and not the entire set of nodes, whereas the leaves would also be well defined clusters of nodes but of a finer granularity.

Just to notice it somewhere, the NWB wiki has a nice list of metrics: https://nwb.slis.indiana.edu/community/ ... a.HomePage

I moved this topic to the newly created "Specifications" section

New (overlapping) community-detection algorithm more than worth a look: Link communities in complex networks. Article published in Nature.

And Edge partitions for overlapping communities in complex networks that I found in the references

Another highly interesting article about clustering, send to me be by Guilhem.

Matthias Bröcheler, Andrea Pugliese, and V.S. Subrahmanian. COSI: Cloud Oriented Subgraph Identification in Massive Social Networks