IUU+DB: LLM system tracks illegal fishing, seafood fraud globally
AI ingests documents to map fisheries crimes and labor abuse across supply chains.
A team of 10 researchers led by Naren Ramakrishnan (Virginia Tech) has introduced IUU+DB, a large language model-based system designed to build a global incident database for illegal, unreported, and unregulated fishing (IUU) plus associated crimes like seafood fraud and labor abuse—collectively termed IUU+. Published on arXiv, the system addresses a critical gap: despite widespread recognition of IUU+ threats to marine ecosystems and livelihoods, quantitative data on their frequency, geography, species, and actors has been scarce.
IUU+DB works by ingesting heterogeneous documents—news articles, scientific reports, court filings—classifying them as relevant incidents, and extracting key data elements such as actors, locations, vessels, species, violations, and enforcement outcomes. It supports deduplication and trend analysis. Validation results show it can surface geographic and behavioral hotspots, assist NGO and academic research, enable source and species risk assessments for industry seafood buyers, and help government agencies target enforcement. The system represents a practical application of LLMs for environmental and social justice monitoring.
- IUU+DB uses LLMs to automatically extract structured data from unstructured documents like news and court records.
- The system covers a broad 'IUU+' scope including illegal fishing, seafood fraud, and labor abuse in fisheries supply chains.
- Case studies validated by researchers show it can identify hotspots and support enforcement, industry risk assessment, and policy implementation.
Why It Matters
LLMs now fight seafood fraud and labor abuse—turning scattered documents into actionable intelligence for regulators and buyers.