AnubisX Framework
A Formal Framework for Behavioral Digital Attribution
16 Formal Axioms
Six groups of logically necessary axioms governing uncertainty, behavior, identity, attribution, evidence, and reasoning.
292 Mathematical Objects
24 categories including 10 formal spaces, 9 distance functions, 7 similarity functions, and a likelihood ratio evidence model.
37 Algorithms
Across 9 domains spanning 5 behavioral modalities: stylometric, chrono-profiling, terminal profiling, network analysis, and media forensics.
4-Tier Validation
31 pre-specified acceptance criteria across unit, component, system, and operational validation tiers with a priori thresholds.
Open Source
Released under CC BY 4.0 with MIT-licensed prototype implementation. DOI: 10.5281/zenodo.21393392.
Cognitive Centroid Theory
A formal model of behavioral identity as an asymptotic attractor in feature space, with four hypothesized formal properties.
Scientific Motivation
Digital attribution \u2014 determining the human source of digital actions \u2014 is a foundational requirement across cybersecurity, digital forensics, and counter-fraud. Current methods rely on technical identifiers (IP addresses, device fingerprints) that sophisticated adversaries can spoof or eliminate. The AnubisX Framework addresses this gap by shifting the analytical focus from transient technical identifiers to persistent human behavioral patterns.
Research Vision
The framework provides: (a) an axiomatic system of 16 axioms governing attribution reasoning; (b) Cognitive Centroid theory \u2014 a formal model of behavioral identity as an asymptotic attractor in feature space; (c) a mathematical framework of 292 defined objects including a likelihood ratio evidence model; (d) 37 specified algorithms spanning five behavioral modalities; and (e) a four-tier validation infrastructure with 31 pre-specified acceptance criteria with thresholds defined a priori.
Six-Layer Architecture
The framework implements a six-layer architecture with a six-stage pipeline supporting identification, verification, and forensic comparison workflows.
Data Ingestion
ImplementedData acquisition through platform-specific adapters
Feature Extraction
ImplementedModality-specific feature computation with normalization
Profile Construction
ImplementedFingerprint generation with quality assessment
Comparison
ImplementedSimilarity computation using configured metrics
Evidence Evaluation
ProposedScore calibration and likelihood ratio computation
Decision
ProposedThreshold-based classification with confidence quantification
Behavioral Modalities
Five behavioral modalities span the framework, each with specified algorithms and implementation status.
Stylometric
Prototype ValidatedLinguistic style analysis, authorship attribution, writeprints
ALG-001 to ALG-004
Chrono-Profiling
SpecifiedTemporal behavior patterns, activity rhythms, posting schedules
ALG-005 to ALG-008
Terminal Profiling
SpecifiedCommand-line patterns, shell behavior, tool usage signatures
ALG-009 to ALG-012
Network Analysis
SpecifiedSocial graph analysis, interaction patterns, community detection
ALG-013 to ALG-016
Media Forensics
SpecifiedImage metadata, editing artifacts, content analysis
ALG-017 to ALG-020
Open Science Commitment
The AnubisX Framework is developed under full open science principles. All components \u2014 axioms, mathematical objects, algorithms, validation criteria, and limitation documentation \u2014 are publicly available for independent validation, replication, and community contribution. The framework explicitly documents all limitations, failure modes, and validation status, establishing a new standard of transparency in behavioral attribution research.
Version 3.0.0 \u2014 Scientific Re-Architecture \u00b7 July 2026 \u00b7 License: CC BY 4.0