Dark Web Threat Actor De-anonymization (SIH26151)
UNVEIL-AI
An investigator-grade dark web intelligence platform built for NTRO to transform fragmented aliases, PGP fingerprints, and cryptocurrency transactions into an explainable, evidence-backed map of threat-actor identities.
0
SIH Problem Statement (SIH26151)
0%
Traceable Evidence Graphs
$0
Cloud Crawling Overhead
The problem
Dark web threat actors operate across fragmented forums, encrypted channels, and cryptocurrency wallets. Existing manual cross-referencing is slow, error-prone, and struggles to establish probabilistic linkage with verifiable evidentiary trails.
Architecture
Graph-first canvas exploration connects disparate dark web entities (crypto wallets, PGP keys, forum personas). Stylometric Persona DNA analysis identifies author fingerprints across posts, while a transparent Noisy-OR probabilistic fusion model provides deterministic, explainable attribution scores.
Hard parts
Correlating multi-platform synthetic personas without active web crawling dependencies, solved by building an extensible entity extraction pipeline and deterministic graph fusion engine with an LLM-assisted investigative narrator.
Where it landed
Functional SIH 2024 platform for problem statement SIH26151 (NTRO), featuring interactive entity graph visualization, automated timeline reconstruction, and human-in-the-loop intelligence dossier export.
Built with
Python · FastAPI · Neo4j · Graph Analytics · LLM Explainer · Stylometry