A Formal Framework for Behavioral Digital Attribution
Version: 3.0.0 — Scientific Re-Architecture
Author: Ahmed Awad (NullC0d3)
ORCID: https://orcid.org/0009-0005-0654-3393
DOI: 10.5281/zenodo.21393392
License: CC BY 4.0
Digital attribution — determining the human source of digital actions and communications — is a foundational requirement across cyber threat intelligence, digital forensics, and counter-fraud. The AnubisX framework provides a formal methodology for behavioral digital attribution.
- Axiomatic Foundation — 16 axioms governing attribution reasoning
- Cognitive Centroid Theory — Formal model of behavioral identity as asymptotic attractor
- Mathematical Framework — 292 objects across 24 categories
- Algorithmic Catalog — 37 algorithms spanning 5 behavioral modalities
- Six-Layer Architecture — Data, Feature, Profile, Comparison, Evidence, Decision
- Four-Tier Validation — 31 pre-specified acceptance criteria
Anubis Twitter v2.5 — stylometric modality for Arabic Twitter data. 47 Python source files (~2,800 LOC). Implements Layers 1-4 of the architecture.
15 proof-of-concept experiments on 31 Egyptian Twitter accounts demonstrate pipeline feasibility:
- 372-dimensional fingerprint extraction: 100% consistency
- FAISS search latency: 6–8 µs
- Cross-user similarity: μ = 0.697, σ = 0.105
- No ground-truth labels available
- 4 of 5 behavioral modalities untested
- Prototype limited to single modality, platform, language
- Not an operational system; research prototype