The repository for our ESMI 2020 Research on calculating Harmonic Centrality on Higher Order Sets
In this project contains methods for converting data files containing hypergraphs into cliques and then transforming the cliques into their adjacency matrix representation. We then model several different networks with their centralization visualized. Finally, we run a couple of statistic calculations to analyze the network's different type of centrality measures.
Graphical Results on the Enron Email Dataset are as follows:
Unweighted Harmonic Centrality Visualization
Weighted Harmonic Centrality Visualization
Unweighted Closeness Centrality Visualization
Analysis Summary on the Enron Email Dataset:
-
Nodes, Edges, and Hyperedges
- Nodes - 143
- Edges - 1800
- Hyperedges - 10551
-
Euclidean Normalization (0-2 scale)
- Unweighted and Weighted Harmonic Centrality - 0.99899
- Unweighted Harmonic Centrality and Unweighted Closeness Centrality - 0.1869
Spearman Correlation Coefficient on Unweighted and Weighted Harmonic Rankings
