Hadith-Inspired Trust Model Detects Hijacking

Post LinkedIn

📄Read original on ArXiv AI

#anomaly-detection#account-hijacking#behavioral-features#trust-modelingmulti-axis-trust-modelingarxivclue-ldscert

💡Interpretable ML framework crushes baselines on account hijacking benchmarks (ROC-AUC 0.715 CERT).

⚡ 30-Second TL;DR

What Changed

Translates 5 Hadith trust axes into 26 user behavioral features

Why It Matters

Provides interpretable alternative to black-box anomaly detection, ideal for securing AI-managed user accounts in cloud environments. Demonstrates robustness on real-world imbalanced datasets, aiding production deployment.

What To Do Next

Extract 26 trust features from your user logs and train a Random Forest model for hijacking detection.

Who should care:Researchers & Academics

Key Points

•Translates 5 Hadith trust axes into 26 user behavioral features
•Introduces temporal features for short-horizon trust changes
•Near-perfect detection on 23k CLUE-LDS windows with Random Forest
•Boosts CERT ROC-AUC from 0.627 to 0.715 with temporal modeling

🧠 Deep Insight

Background and context from public sources — not the original article. 5 sources cited.

🔑 Enhanced Key Takeaways

•The paper was submitted to arXiv on February 20, 2026, by author Mohammad AL-Smadi[1].
•Five specific Hadith trust axes translated are long-term integrity (adalah), behavioral precision (dabt), contextual continuity (isnad), cumulative reputation, and anomaly evidence[1].
•On a leakage-controlled 4,000-user CERT r6.2 configuration, temporal modeling improves PR-AUC from 0.072 to 0.264 beyond ROC-AUC gains[1].

🔮 Future ImplicationsAI analysis grounded in cited sources

Hadith-inspired models will appear in 10% more cybersecurity papers by 2028

Interpretable multi-axis frameworks outperform black-box baselines on standard datasets like CERT, encouraging academic adoption of culturally inspired trust modeling.

Temporal trust features will boost ROC-AUC by 0.05+ in 70% of imbalanced hijacking benchmarks

Paper demonstrates consistent gains from lightweight temporal signals on sparse malicious behavior in CERT r6.2, applicable to real-world enterprise logs.

⏳ Timeline

2026-02

Paper submitted to arXiv as v1 by Mohammad AL-Smadi

📎 Sources (5)

Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.

📄Read original article on ArXiv AI

📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

Same topic

Explore #anomaly-detection

Same product