HACKING PROJECT · social · Advanced
Disinformation Narrative Detection & Mapping Dashboard
Automatically cluster, track, and visualize disinformation narratives and their propagation networks across platforms.
Coordinated disinformation campaigns spread false narratives across platforms faster than fact-checkers and analysts can track them, and manual monitoring cannot keep pace with volume or cross-platform diffusion. Analysts lack tooling that clusters related claims into coherent narratives, tracks their evolution and spread, and links them to amplifier networks. This blind spot undermines election integrity, public health, and crisis response.
**What you'll build**
Build a NLP-driven pipeline that ingests multi-source content, clusters posts into narratives via embeddings and topic modeling, detects claim variants and framing, and maps temporal/geographic spread plus amplifier networks on an interactive dashboard. Integrate stance/claim detection and align findings to the DISARM disinformation framework so narratives are tagged with adversary TTPs and tracked over time.
**Objective** — Automatically cluster, track, and visualize disinformation narratives and their propagation networks across platforms.
**Expected outcome** — An interactive dashboard that surfaces active narratives, their spread timeline/network graph, and DISARM-tagged campaign profiles exportable as intel reports.
**Difficulty** — Advanced — combines NLP (embeddings, clustering, stance detection), graph analytics, and geopolitical/framework analysis.
**Primary category** — Threat Intelligence / Influence-Ops Analytics; Secondary = OSINT/SOCMINT, AI/ML Security, Data Science
**Security domains** — Primary = Threat Intelligence; Secondary = Security Research, AI Security, Blue Team
**Technical focus** — Narrative clustering, topic modeling, semantic embeddings, claim/stance detection, temporal diffusion analysis, social network analysis, geospatial mapping, coordinated-behavior linkage, data visualization
**Core technologies** — Languages: Python, JavaScript; Frameworks/Libraries: sentence-transformers, BERTopic/HDBSCAN/UMAP, spaCy, NetworkX, D3.js/Plotly/Kepler.gl; Security Tools: Maltego, DISARM framework tooling; Cloud/Infra: Docker, Kafka (streaming), cloud storage; AI/Data: LLM for claim summarization, vector DB; Databases: Neo4j, PostgreSQL, Elasticsearch, vector store (FAISS/Milvus)
**Where it's valued** — Elections integrity/civil society, government/defense, threat-intel firms, journalism/fact-checking, brand & reputation management, public health
**Careers it builds toward** — Disinformation/Influence-Ops Analyst, Threat Intelligence Analyst, OSINT Researcher, Data Scientist (Integrity), Intelligence Analyst
**Framework alignment**
• MITRE ATT&CK: Use DISARM (AMITT) framework as the primary TTP taxonomy (e.g., Develop Narratives, Seed Distortion, Amplify, Cross-Platform Spread). ATT&CK Enterprise is N/A for the influence layer.
• NIST CSF 2.0: Identify (threat landscape mapping), Detect (narrative/anomaly detection), Respond (intel product generation)
• OWASP: LLM Top 10 — LLM05/LLM09 awareness (misinformation) if using LLM summarization; otherwise N/A
• CIS Controls v8: Largely N/A; Control 8 (audit/provenance of collected data) applies
**Innovation** — Research-Level Project — narrative mapping at scale with DISARM alignment is at the frontier of academic and vendor R&D and has clear productization upside. Scores: Innovation 8/10, Startup 7/10, Enterprise-Readiness 6/10, Patent = Possible.