BEMO API v1 · ONTOLOGY v0.1.0

Developer documentation

The public API exposes 488 complete BEMO metric records, 488 computation profiles, 1,392 source-declared relationships, and the separate 2,185-record discovery catalog. Only 62 profiles expose a generic proportion or ratio template; all other calculation requests are refused until their required scientific protocol is registered and validated.

Metric search

GET https://evidencemetrics.org/api/v1/metrics/?query=specificity&implementation=TemplateComputable&limit=20

Parameters: query, category, type (scientific pillar), status (lifecycle), implementation (readiness), validation_status (formula status), target (scale), sort, page, and limit.

Metric detail

GET https://evidencemetrics.org/api/v1/metrics/BEMO-2000003

Returns the complete scientific record, computation profile, source record, references, outgoing relationships, and incoming relationships.

Generic calculation template

POST https://evidencemetrics.org/api/v1/calculations/
Content-Type: application/json

{
  "metric_id": "BEMO:2000003",
  "metric_version": "0.1.0",
  "input": {
    "inputs": {"numerator": 92, "denominator": 100},
    "uncertainty": {"method": "prespecified protocol"},
    "evidence_snapshot_id": "urn:example:snapshot:001",
    "assessed_object": {"id": "urn:example:object:001", "type": "evidence-set"}
  }
}

Only TemplateComputable profiles with the declared divide expression run. Requests require uncertainty, an immutable evidence-snapshot identifier, and an assessed object.

Refusal behavior

POST /api/v1/calculations/\nmetric_id: BEMO:2000001

The service returns 422 protocol_required for context-specific, rubric, model, assay, or operational-definition profiles. It never substitutes a generic score.

Document evaluation

POST https://evidencemetrics.org/api/v1/document-evaluations/
Content-Type: multipart/form-data

document=@paper.pdf
study_type=clinical_trial
primary_claim=The intervention improves the primary outcome.

Uploads are limited to 10 MB PDF, DOCX, TXT, Markdown, or RTF documents. The AI proposes 8-20 relevant released BEMO metrics with short evidence excerpts and locations. EvidenceMetrics does not retain the document or result, and all values remain ai_proposed until human review.

Document-evaluation safeguards

GET /api/v1/document-evaluations/

Returns configuration status without exposing credentials. Same-origin checks and privacy-preserving quotas protect POST requests. Only registered generic divide templates are calculated locally; context-specific metrics return protocol_required.

Discovery catalog

GET https://evidencemetrics.org/api/v1/catalog/?query=calibration&limit=25

The separate source-only Part 11 discovery catalog remains available with explicit missing-specification disclosure.

Research use only. Technical execution and structural validation do not establish scientific validity or authorize patient-specific decision-making.