[Networkit] Introducing Network Profiles
christian.staudt at kit.edu
Wed Sep 9 16:35:54 CEST 2015
I would like to introduce you to a new feature coming with NetworKit 4.0.
The aim of the tool suite is to make it easy to characterize the structure of a large complex network, e.g. in order to link structural properties and various aspects of the complex system modelled by the network. The new profiling module provides an easy way for exploratory network analysis. Previously, properties.overview has been a makeshift version of that, but will be retired in favor of something much more systematic and extensive.
Here are a couple of example profiles:
a connectome network, mapping physical connections in the human brain
a Facebook friendship network of a US women’s college
a hyperlink web graph of the .uk domain
Besides general information about the graph, we calculate various node centrality measures and characterize/visualize their distributions. Then correlations between these measures are computed and visualized. We also look at ways to partition the network (components, communities, cores and shells) and plot the resulting subset size distributions. We are open for the inclusion of more features.
Before releasing this, I am asking for feedback and help with beta testing this new feature. If you have any suggestions, feel free to let me know or discuss them here. To try it out, get the current Dev branch and run:
G = …. # some network
profile = profiling.Profile.create(G)
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
More information about the NetworKit