Fabled Sky Research

AIO Standards & Frameworks

AIO Scoring Framework: Tier Definitions

FSRAIOAPI
December 11, 2022

Document Type: Framework
Section: Docs
Repository: https://aio.fabledsky.com
Maintainer: Fabled Sky Research
Last updated: April 2025

Overview

The AIO Scoring Framework provides a normalized, tier-based rubric for evaluating the information quality and machine-actionability of any digital asset audited under Artificial Intelligence Optimization (AIO) standards. This document defines three tiers—Compliant, Optimized, and Preferred—and details how the four AIO pillars (Retrievability, Metadata, Structure, Traceability) are individually measured and collectively weighted to generate a deterministic score.

Scope

• Applies to webpages, PDFs, datasets, and structured data evaluated with the official AIO Audit Kit (v2.3 ≥).
• Consumable by auditors, SEO strategists, developers, content teams, and automated pipelines.
• Outputs a numeric score (0-100) plus descriptive tier.

Definitions

Retrievability (R) – The ease with which humans and algorithms can locate the asset using standard discovery mechanisms (indexing, crawling, site search, external search).
Metadata (M) – Accuracy, completeness, and standards-compliance of descriptive data (Schema.org, Dublin Core, HTTP headers).
Structure (S) – Semantic and syntactic organization of content (HTML5 outline, heading hierarchy, ARIA roles, CSV/JSON schema validity).
Traceability (T) – Degree to which claims, facts, and data elements are source-linked, version-controlled, and timestamped for provenance.

Pillar Weighting Model

The framework uses a constant weight distribution across all tiers:

Retrievability = 30%
Metadata       = 25%
Structure      = 25%
Traceability   = 20%

Each pillar is scored on a 0–5 scale by the Audit Kit. The raw pillar score is multiplied by its weight to produce a weighted pillar score (WPS).

Weighted Pillar Score (WPS) = (PillarScore / 5) * PillarWeight * 100
Total Score (TS)            = Σ WPS  // Range: 0–100

Tier Threshold Matrix

Tier	Total Score (TS)	Minimum Pillar Floor*
Compliant	60–79	50% of pillar weight
Optimized	80–89	70% of pillar weight
Preferred	90–100	90% of pillar weight

*Pillar Floor: A content asset cannot qualify for a tier if any individual pillar falls below the stated floor, even if TS meets the threshold.

Pillar-Specific Measurement Criteria

Retrievability

• Robots.txt returns 200 and allows crawling.
• Canonical and hreflang tags valid.
• ≤2 redirects to reach canonical URL.
• Asset is indexed by at least one major search engine within 72 h of submission.

Metadata

• Valid JSON-LD block with @context: "https://schema.org" and primary @type.
• Mandatory fields (name, description, inLanguage, datePublished) present.
• Open Graph and Twitter Card parity.
• HTTP Last-Modified header properly set (RFC 7232).

Structure

• Single `

` followed by descending heading levels; no skipped levels. • Landmarks (`
`, `
`, `
`) present. • Data tables include ` `, ``, ``. • PDFs tagged (PDF/UA-1). #### Traceability • All factual claims reference a verifiable source with persistent identifier (DOI, URI, or hash). • Version history accessible via `changelog.json`. • Timestamp granularity ≤1 h UTC. • Digital signatures for datasets (SHA-256). ### Implementation Workflow 1. Run `aio-audit –url –profile standard` to collect pillar metrics. 2. Audit Kit returns JSON payload (see Example Audit Output). 3. Internal or CI pipeline computes TS and cross-checks pillar floors. 4. Tier label is assigned programmatically; stored in `aio_score.json`. 5. Optional: push results to centralized AIO Registry for benchmarking. ### Example Audit Output (JSON) “`json { “@context”: “https://aio.fabledsky.com/v2”, “auditedAt”: “2025-04-17T11:26:40Z”, “target”: “https://example.com/research-paper”, “pillarScores”: { “retrievability”: 4.5, “metadata”: 4.2, “structure”: 4.0, “traceability”: 3.8 }, “weightedPillarScores”: { “retrievability”: 27.0, “metadata”: 21.0, “structure”: 20.0, “traceability”: 15.2 }, “totalScore”: 83.2, “tier”: “Optimized”, “version”: “AIO-SF-1.0” } “` ### Best-Practice Checklist by Tier • Compliant → Verify robots rules, supply minimal JSON-LD, add canonical. • Optimized → Implement full Schema.org type, maintain heading hierarchy, use changelog. • Preferred → Add source-hash ledger, automated sitemap ping, continuous monitoring hooks. ### Continuous Monitoring Assets graded Preferred must rerun audits every 30 days; Optimized every 90 days; Compliant on major revision. Use the `–watch` flag to trigger deltas and maintain tier status. The AIO Scoring Framework ensures that digital assets adhere to transparent, machine-verifiable standards while incentivizing progressive enhancement toward full Preferred status. Consistent application of this rubric supports reproducible audits, higher search discoverability, and robust provenance across the AI supply chain.