The latent cognitive structures of social networks

A jupyter notebook tutorial corresponding to the work The latent cognitive structures of social networks by Izabel Aguiar and Johan Ugander (2023).

The CSS NNTuck Code Tutorial.pynb notebook steps through the methods discussed and performs the NNTuck on the Krackhardt Advice CSS from Krackhardt (1987).

The code depends on the following packages and the version number for which it is reproducible is noted. numpy (version 1.22.2), tensorly (version 0.5.1), sklearn (version 0.23.2), networkx (version 3.0), and matplotlib (version 3.3.2). When sweeping over parameters $K$ and $C$ in the NNTuck it is most efficient to run the sweep in parallel, which depends on joblib (version 1.0.1) and on os to make sure the parallel runs don't use too much CPU.

Any additional tools which may not be descibed here are described in the github page for A factor model of multilayer network interdependence (Aguiar, Taylor, Ugander, 2022).

If you use this code, please cite The latent cognitive structures of social networks (Aguiar and Ugander, 2023) and/or A factor model of multilayer network interdependence (Aguiar, Taylor, Ugander, 2022)..

If you have any questions about this code repository or the work in general, please email me at izzya@stanford.edu!