How the dataset is defined, collected, classified and de-duplicated
This methodology documents the steps taken to build a dataset of emerging ocean economy companies in Florida. This project is a personal exercise in learning about the ocean economy landscape in Florida. Using publicly available sources and with the assistance of Claude, I have scraped online sources for companies, categorized them by sector and innovation theme, and compiled summary information on each one. The goal will be to continue bringing in new data sources as I find them, and to explore ways to make this database useful to the Florida community.
The methodology is laid out in the following sections:
No single source captures the ocean economy. Companies surface through different channels at different stages — federal grant databases catch R&D-stage firms, while local accelerators and award programs catch ventures earlier, often before they have any funding record at all. The dataset therefore combines multiple discovery channels, with each new source parsed into the same standardized company-level schema so that everything can be merged, deduplicated, and compared.
The U.S. federal government offers Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) awards to emerging startups. Data on SBIR/STTR awards is published on a quarterly basis, and includes information such as award abstracts, research keywords, and activity descriptions that make the dataset convenient to analyze.
Search process: the full national dataset is downloaded from SBIR.gov via its API. The following filters were applied:
| Federal Agency | Included Branches |
|---|---|
| Department of Agriculture (USDA) | Incl. National Institute of Food and Agriculture (NIFA) |
| Department of Commerce (DOC) | National Oceanic and Atmospheric Administration (NOAA); National Institute of Standards and Technology (NIST) |
| Department of Defense (DOD) | Incl. Navy and Defense Advanced Research Projects Agency (DARPA) |
| Department of Energy (DOE) | Incl. ARPA-E |
| Department of Homeland Security (DHS) | Science and Technology Directorate |
| Department of Transportation (DOT) | All branches |
| Environmental Protection Agency (EPA) | All branches |
| National Science Foundation (NSF) | All branches |
Several start-up accelerators and incubators operate in Florida, including Seaworthy Collective and Ocean Exchange. Each organization publishes information about the startups that it has supported. Web searches were conducted to scrape information from these organizations.
Additional programs are added as they are identified — including The Continuum, a NOAA-funded national network of ocean-economy accelerators (Tampa Bay Wave's BlueTech|X and Braid Theory cohorts, among others), and FAU Tech Runway's pitch-competition awards. Cohort lists, award announcements, and alumni pages are parsed into the standard schema, with a source URL retained for every record.
Companies are classified using Anthropic's Claude AI. Classification is driven by the source text available for each company — award abstracts and research keywords for federally funded companies, program descriptions and cohort announcements for accelerator-sourced companies. Where the source text is ambiguous, the company website is reviewed for evidence of the company's primary line of business.
Why AI classification?
Each company is classified along two dimensions: ocean economy sector and innovation theme. The Sector reflects the established industry in which the company operates; the Innovation Theme reflects the emerging blue-economy activity it represents. A company can sit in an established sector and carry an emerging theme — an eco-engineered seawall venture, for example, is Construction sector + Coastal Resilience theme.
Sectors are aligned to FAU's Florida ocean-economy industry framework, keeping the dataset comparable to FAU's published economic baseline for the state.
| Category | Description |
|---|---|
| Construction | Coastal protection and resilient marine infrastructure — eco-engineered shorelines, 3D-printed coastal structures, flood barriers |
| Energy | Marine renewable energy — wave energy converters and grid-scale marine energy systems |
| Living Resources | Aquaculture, fisheries, and marine biotechnology — broodstock systems, species identification, sea-lice removal |
| Marine Industry | Vessel operations and maintenance — hull grooming, cleaning robotics, and shipboard systems |
| Military | Naval and undersea defense technology — sonar, undersea sensing, and ship systems |
| Ocean Intelligence | Ocean observation and data — autonomous underwater vehicles, in-situ sensor networks, underwater LiDAR |
Established industry frameworks report jobs and GDP but have no home for emerging blue tech such as blue carbon, marine-plastics remediation, or ecosystem restoration. The Innovation Theme layer surfaces these emerging niches — exactly the untracked segments this project targets. Ventures with no established industry home (e.g., blue-carbon fintech) are tagged Cross-Sector / Enabling Tech at the sector level, and their Innovation Theme carries the real signal.
| Theme | Description |
|---|---|
| Ocean Data & Intelligence | Ocean sensing, observation, autonomous platforms, and marine data & analytics |
| Sustainable Materials | Bio-based, recycled, and ocean-safe alternatives to conventional materials |
| Ecosystem Restoration & Conservation | Coral, reef, and coastal habitat restoration and conservation technology |
| Regenerative Ocean Food | Sustainable aquaculture, mariculture, and regenerative seafood systems |
| Coastal Resilience & Adaptation | Flood protection, living shorelines, and climate-adaptation infrastructure |
| Ocean Carbon / Blue Carbon | Ocean-based carbon removal, measurement, and blue-carbon markets |
| Pollution, Plastics & Marine Debris | Detection, prevention, and removal of marine pollution and debris |
Source records are often incomplete — accelerator cohort pages may list only a company name, and federal award records can carry outdated contact details. AI-assisted web research is used to fill these gaps for each company:
AI classification and enrichment are a first pass, not the final word. Spot checks were conducted across the dataset to verify the accuracy of company descriptions and the validity of website links, with corrections applied directly to the affected records.
Every company in the dataset is also reviewed to verify ocean relevance, using the company website and other publicly available information. Companies appearing in multiple sources — a Seaworthy Collective graduate that later wins a federal award, for example — are matched to a single master record rather than counted twice. Misclassifications identified in review are corrected in the dataset, and review notes are retained alongside each record.
Sources are refreshed on a periodic basis: SBIR.gov quarterly as new awards are published, and organization rosters as new cohorts and award cycles are announced. The current dataset includes companies identified through July 2026. Classification is re-run when new data is added to ensure consistency across the full dataset, the ocean-relevance review is repeated for newly added companies, and cross-source deduplication is re-applied before new records enter the master dataset.