In the bad old days, we developed functionality (in temsna) for doing all the network-based similarity comparisons, e.g. see "Structural Diversity is a Poor Proxy for Information Diversity." We want to be able to easily generate lists of nodes that interact / connect under specific conditions (e.g., ego+alter, ego+alter+alter-alter, etc.). We did this with networkx in the past, but now we're a graph-tool shop. Re-implement the older stuff here in graph-tool, and expand on that functionality as needed given differences in our data, project goals, etc.
Here's the key image from the Social Networks paper that illustrates some of the different network neighbourhood comparisons we worked on in past projects. (Note that we don't necessarily need / want to implement everything as we did before! Focus on what makes sense for this project.)
In the bad old days, we developed functionality (in
temsna) for doing all the network-based similarity comparisons, e.g. see "Structural Diversity is a Poor Proxy for Information Diversity." We want to be able to easily generate lists of nodes that interact / connect under specific conditions (e.g., ego+alter, ego+alter+alter-alter, etc.). We did this withnetworkxin the past, but now we're agraph-toolshop. Re-implement the older stuff here ingraph-tool, and expand on that functionality as needed given differences in our data, project goals, etc.Here's the key image from the Social Networks paper that illustrates some of the different network neighbourhood comparisons we worked on in past projects. (Note that we don't necessarily need / want to implement everything as we did before! Focus on what makes sense for this project.)