A family of open, reproducible workflows, that transform raw data into structured, multi-layered biodiversity knowledge products you see on this portal.
The problem it solves. Converting occurrence records into consistent, reusable knowledge products typically relies on ad-hoc manual pipelines. cheCkOVER automates the entire transformation.
Taxonomy & origin. Harmonizes taxonomy, retains type locality and vernacular names, and assigns population origin (indigenous vs non-indigenous) — anchoring every species to its nomenclatural identity.
Spatial metrics. Derives IUCN-aligned EOO and AOO, then intersects occurrences with administrative units, hydrographic basins, ecoregions, and protected areas for full geospatial contextualization.
AI- & API-ready outputs. Delivers GeoJSON, KML/KMZ, JSON, and Markdown exports via REST endpoints, plus a canonical geo-narrative structured for machine consumption — including RAG/LLM pipelines.
Time-aware mapping. Local extinction is not simply absence — it is a directional, high-information transition (presence → loss). Where extinction-claim records exist, cheCkOVER flags distributional change signals and prevents temporal inflation in IUCN range metrics.
Evidence manifestation & SEBs. Artifacts are packaged as versioned, citable Species Exposure Bundles (SEBs) — machine-readable artefacts with explicit provenance. LLM tests showed manifest-directed SEB retrieval reached 6.97/8 vs 1.78/8 closed-book, with 0% failures.
Community Data & tiered access. cheCkOVER preserves contributor-defined confidentiality tiers at the raw-data level, while derived outputs never expose precise coordinates—keeping restricted records protected.
Proven at scale. Demonstrated on 465 crayfish species across 97 countries — from 2 to over 20,000 records per species — showing robust operation across extreme heterogeneity in range size and record density.