Graph Robustness Score C(G) – live interactive demo with spectral gap + Ollivier-Ricci + attack simulation (feedback welcome!)

Show the Community!
1

Hey Streamlit community! :waving_hand:

I just built Graph Robustness Score — a one-click tool that gives any network one clean scalar C(G) telling you how robust it is.

What it does:

  • Upload an edgelist (or try sample graphs)
  • Adjust weights for 3 complementary metrics:
    • Spectral gap (global connectivity)
    • Singular-value variance (hub fragility penalty)
    • Real Ollivier-Ricci curvature (local geometry)
  • Instantly see C(G) + before/after attack simulation (random vs targeted hub removal)

Live demo (no login, no install):

https://graph-robustness-score.streamlit.app

GitHub repo (full code + paper + testbed):

https://github.com/syedrazaaftab/graph-robustness-score

Here’s what the app looks like:

IMG_4838
IMG_48381284×980 81.2 KB

Would love your honest feedback on:

  1. UI/UX – is it clear and useful?
  2. The combined metric – does blending spectral + singular-value + curvature make sense?
  3. Performance on larger graphs
  4. Any features you’d want next (e.g. more sample graphs, export report, etc.)

Thank you! Excited to hear what you think :rocket: