The same habitat has three different names depending on who is describing it. HabitatMech resolves them onto one ontology-grounded record and keeps every source's own account of it.
| Source | How it names marine sediment |
|---|---|
| JGI GOLD | Environmental > Aquatic > Marine > Sediment |
| BacDive | Marine-sediment |
| PREGO | ENVO:00002113 |
HabitatMech is not a re-labelling of ENVO. Most of what it holds, ENVO has no term for — and a concept that merely hangs under an ENVO term counts as new, because the identifier is minted precisely when the matched term is broader than the concept.
| Records | Assertions | Status |
|---|---|---|
| 532 | 373065 | ENVO names this concept directly |
| 626 | 69464 | another vendored OBO ontology names it; ENVO does not |
| 666 | 97166 | no term, but narrower than an ENVO term it hangs under |
| 689 | 41920 | no term, but narrower than a non-ENVO term |
| 717 | 72186 | no ontology term names it or anything above it |
| Category | Records | Share |
|---|---|---|
| HOST_ASSOCIATED | 1642 | |
| ENGINEERED | 486 | |
| AQUATIC | 469 | |
| TERRESTRIAL | 353 | |
| OTHER | 208 | |
| FOOD | 61 | |
| AIR | 7 | |
| CLINICAL | 4 |
Every record says how well its identifier fits the source concept, separately from whether a human has checked it. A record can be reviewed and still honestly ungrounded — that combination is a well-formed request for a new ontology term, and there are 118 of them.
| Status | Records | Meaning |
|---|---|---|
| EXACT | 1062 | The identifier is an exact match for the source concept. |
| NARROW | 987 | The identifier is narrower than the source concept. |
| UNGROUNDED | 981 | No defensible ontology term; a minted identifier and a term-request candidate. |
| NOT_APPLICABLE | 97 | Not a habitat at all — a host taxon, a disease process, a quality. |
| CLOSE | 90 | Close, but not exact and not cleanly broader or narrower. |
| BROAD | 13 | The identifier is broader — the best available term, losing specificity. |
This corpus is machine-generated. 2569 of 3230 records have had no human review. The harmonization rules are documented and tested, but a lexical match between a source label and an ontology label is a plausible grounding, not a verified one.