Research map / Learning theory and algorithms / Combinatorial and search optimization
Hide Exposures by Removing Mastermind’s External Sources on Social Network (Student Abstract)
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes HERMES, a heuristic that identifies a mastermind in a core community via centrality ranking and rewires its external links to core members to hide it from influence-based detection, outperforming ROAM on Facebook and Google networks.
Problem
Hiding the mastermind of a covert network on social media so that investigators cannot trace it through its external connections, while keeping the algorithm efficient.
Approach
Louvain community detection plus DC/CC/BC rank aggregation locates the core community and mastermind, then iteratively removes the mastermind's most influential external links and reconnects those nodes to less influential core members, rather than operating on the whole network as ROAM does.
Key results
- HERMES reduces time to identify the mastermind by 103 seconds on Facebook (220 vs 323 sec) and 1397 seconds on Google (4387 vs 5784 sec) compared with ROAM.
- HERMES achieves an LT influence value up to 11.11% higher than ROAM on Facebook, while IC values are lower by about 1.4%.
- On Google, HERMES yields IC values 1.92% to 4.00% lower and LT values 2.63% to 2.85% higher than ROAM across iterations.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use HERMES instead of ROAM | Mastermind hiding on Facebook network · Facebook | time: 323 sec -> 220 sec | Helped |
| Use HERMES instead of ROAM | Mastermind hiding on Google network · Google | time: 5784 sec -> 4387 sec | Helped |
| Use HERMES instead of ROAM | Mastermind hiding on Facebook network · Facebook | IC: 0.72 -> 0.71; 0.71 -> 0.70; 0.70 -> 0.69 | Mixed |
| Use HERMES instead of ROAM | Mastermind hiding on Facebook network · Facebook | LT: 0.09 -> 0.10; 0.11 -> 0.11; 0.12 -> 0.13 | Mixed |
| Use HERMES instead of ROAM | Mastermind hiding on Google network · Google | IC: 0.52 -> 0.51; 0.51 -> 0.50; 0.50 -> 0.48 | Hurt |
| Use HERMES instead of ROAM | Mastermind hiding on Google network · Google | LT: 0.35 -> 0.36; 0.37 -> 0.38; 0.38 -> 0.39 | Helped |
Limitations
- Evaluated only on two networks and a single illustrative 9/11 example in the text.
- Results are mixed: IC value is not lower than ROAM in iteration 2 on Facebook, and the Google time gain is reported without variance.
Experimental setup
Datasets: Facebook network (4039 nodes, 88234 edges) Google network (15763 nodes, 149456 edges) 9/11 terrorist network
Benchmarks: Facebook network Google network
Metrics: Independent Cascade influence value Linear Threshold influence value runtime in seconds
Baselines: ROAM
Builds on: ROAM Louvain community detection
Closest papers by meaning
- Hiding in Multilayer Networks (AAAI 2020 · similarity 0.81)
- Deception through Half-Truths (AAAI 2020 · similarity 0.78)
- Eliminating Majority Illusion Is Easy (AAAI 2025 · similarity 0.77)
- Adversarial Socialbots Modeling Based on Structural Information Principles (AAAI 2024 · similarity 0.76)
- Differentially Private Graph Diffusion with Applications in Personalized PageRanks (NeurIPS 2024 · similarity 0.75)
- Harnessing Network Effect for Fake News Mitigation: Selecting Debunkers via Self-Imitation Learning (AAAI 2024 · similarity 0.75)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.