21KDD-Momo

Folders

Folder	Description
alignments	Gigi alignments computed without node alignments
alignments-jaccard	Gigi alignments computed with node alignments
code	Code for Beppo, Gigi, NMD computation, plotting, and table generation
data	Preprocessed input data as used in all experiments
graphics	Figures from paper and further visualizations
network-portrait-divergence	NPD code and NPD heatmap
raw-data	Raw input data as provided by the stated sources (if different from the preprocessed data)
results	Beppo results with Julia IDs
results_mapped	Beppo results with original IDs

All folders have a separate README.md that describes their contents or usage.

Setup

The code runs on Debian Debian GNU/Linux 10 (buster) and macOS Catalina; other operating systems were not tested.

We need the following packages when running with Julia version 1.0.3 (this also works with Julia 1.3.0 and slightly older versions of some packages):

       "ArgParse" => v"1.1.0"
            "CSV" => v"0.7.4"
     "DataFrames" => v"0.21.4"
 "DataStructures" => v"0.17.19"
        "GraphIO" => v"0.5.0"
      "GraphPlot" => v"0.4.2"
           "JSON" => v"0.21.0"
    "LightGraphs" => v"1.3.3"
   "LineSearches" => v"7.0.1"
          "Optim" => v"0.22.0"
      "StatsBase" => v"0.33.0"
           "YAML" => v"0.4.0"

This is the package setup with which the results showcased in the paper were generated. A better practice for distributing Julia code, which we only learned about later, is documented here. For convenience, we also included requirements-minimal Manifest.toml and Project.toml files that should work, e.g., with Julia 1.6, in this folder.

Julia allows multiple versions of itself to coexist, i.e., we can download Julia 1.6 and set up an alias like julia16 in the startup script of our preferred shell (e.g., ~/.zshrc), even if we are normally working with a different version.

To run the code using the more professional setup (assuming Julia 1.6), do the following:

1. Activate the julia16 REPL
2. Type ] to go to the package REPL
3. Type activate . (this should cause the string between the round brackets to the left of your prompt to change)
4. Type instantiate (this will install all packages in the versions required by Manifest.toml and allow you to run Julia code)

To run the Python code:

1. Set up a virtual environment with Python 3.7 and activate it (recommended, not required)
2. Run pip install -r requirements.txt

Detailed instructions on how to reproduce the results are given in the README.md of the code folder.