I’m facing a situation in which I have a network of more than 19k nodes, but I need the extended clustering coefficient to be evaluated only for around 600 nodes at a time. In fact I’m iterating a function that makes changes to the graph each time and gets the coefficient only for some nodes.
I need to reduce as much as I can the execution time for the coefficient evaluation and to do so I tried to build up a mechanism that considers ego-network (max distance = 2) for each needed node and computes the coefficient, but since most of the nodes are pretty much near each other it doesn’t solve the problem.
Did someone have to deal with such situation? Any suggestion on possible solutions?
Adding an option in the function that lets you consider a subset of nodes would be a lot useful, but I can imagine there may be more relevant features to add.