About ZebulunAI
ZebulunAI was founded from an idea on a paper entitled "Organised Crime Analysis Using NLP and Sentiment Analysis from SMS" which focused on SMS‑based NLP for law‑enforcement investigations. Our research addresses the challenges of slang‑rich, noisy mobile data and demonstrated how to normalise language, extract entities, and map relationships to surface actionable leads. We are turning that work into secure, operational software for investigators and analysts.
Slang → Standard Translation
Neural translation that converts slang‑filled SMS/social text into clear English while preserving intent—improving search, triage, and downstream NLP accuracy.
Areas of Interest Highlighter
Ranks segments in long transcripts and recordings, surfacing people, places, and topics so reviewers can jump to the most relevant parts first.
Balance of Power (BoP)
Graph‑aware models that infer hierarchy and roles within organised‑crime communications from message flow and linguistic cues.
Linguistic Fingerprinting
Idiolect features that help link numbers and aliases to the same speaker across datasets—with confidence scores and privacy controls.
Our Principles
- Safety & Ethics: Human‑in‑the‑loop by default, bias checks, and transparent model behaviour suitable for audit.
- Security: Least‑privilege access, encryption in transit and at rest, and comprehensive logging.
- Interoperability: Standards‑first integrations with existing systems and data sources.
- Delivery Discipline: Clear milestones, reproducible experiments, and measurable outcomes.