1. Weighted Scoring Model
Every ResourceRank composite score ($S \in [0, 100]$) is computed as a weighted sum across four orthogonal evaluation dimensions. Each dimension must be backed by verifiable evidence artifacts.
Provenance & Lineage
Verifiable origin, authorship, licensing clarity, version history, and supply-chain integrity.
Required Evidence Indicators
- Cryptographically signed releases or verifiable checksums
- Explicit SPDX license identifiers and contributor attribution
- Transparent changelog and historical versioning
Empirical Evidence & Reproducibility
Quantitative benchmarks, reproducible methodologies, independent audits, and peer-reviewed validation.
Required Evidence Indicators
- Reproducible test suites or benchmark harnesses
- Independent third-party security or quality audits
- Documented methodology with raw data availability
Operational Reliability & Maintenance
Active stewardship, issue resolution velocity, architectural resilience, and backward compatibility.
Required Evidence Indicators
- Sustained commit cadence and multi-maintainer governance
- Automated CI/CD verification and regression coverage
- Documented deprecation policies and SLA transparency
Open Standards & Interoperability
Adherence to open protocols, portable data formats, non-proprietary APIs, and low exit friction.
Required Evidence Indicators
- Support for open schemas (JSON-LD, Parquet, OpenAPI, etc.)
- Complete data export without proprietary lock-in
- Composable integration primitives
Classification Thresholds
90–100: Meets all provenance, reproducibility, and open standard criteria with independent verification.
75–89: Production-ready resource with documented lineage and minor governance or portability caveats.
60–74: Usable with explicit mitigations; incomplete audit trails or partial lock-in risks identified.
0–59: Insufficient evidence, unclear licensing, or unverified provenance; not recommended for critical adoption.
Composite Score Simulator
Test how dimension scores combine under the ResourceRank v1.0 weighting specification.
(90×0.25) + (85×0.30) + (88×0.25) + (92×0.20) = 88.4
2. Provenance & Evidence Artifact Rules
Evidence Verifiability
Claims without verifiable artifacts are scored at zero for the corresponding indicator. Acceptable artifacts include:
- Cryptographic hashes (SHA-256 / SHA-512) and Sigstore attestations
- Persistent DOIs for benchmark papers and reproducibility packages
- Publicly archived governance votes and SPDX license manifests
- Reproducible containerized evaluation harnesses
Decision Record Linkage
Every evaluation should link to at least one ResourceRank Decision Record (RDR) documenting why a resource was selected, conditionally approved, or rejected for a concrete architectural or research context.
Inspect Decision Records