Venue Ranking Evidence

Venue Ranking Evidence

Resolve venues, report source-specific observations, and preserve official-page proof. Use the runtime for deterministic matching, provenance, and artifacts; use judgment only to explain ambiguity and provider limitations.

Live paths: ICORE (conference ranks, edition discovery, optional browser proof) and DOAJ (public journal membership CSV; live or import; always currentness-unconfirmed; proof-ineligible). CCF, SCImago, Scopus, Web of Science Master Journal List, JCR, JUFO, the Norwegian Register, and Conference Ranks accept authorized normalized CSV/JSON imports only. Imports remain currentness-unconfirmed. Conference Ranks is secondary-legacy and must never be presented as latest.

Required workflow

1. Run doctor when runtime or browser readiness is unknown.
2. Run sources list|show|check to confirm capabilities, optional --data-file preflight, and deferred-source notes.
3. Run lookup with the user's text and requested sources. Inspect the delivery matrix (status, match_status, source_coverage_status, incomplete_analysis, requested_sources / satisfied_sources / missing_or_blocked). Never treat ready as multi-source completeness unless every requested source is satisfied.
4. Separate observations by source, assertion kind, scheme, category or collection, and metric year or edition. Never summarize a venue as merely “Q1”, “ranked”, “Scopus”, or “WoS”.
5. Unique exact-identifier / exact-title / exact-alias identity resolves even when weaker fuzzy candidates remain; fuzzy-only short acronyms stay ambiguous. Full candidates remain in matches.jsonl (display may be top-K).
6. If proof is requested and identity is ambiguous, obtain an explicit --venue-id selection first. Unique exact-tier resolution may auto-select for proof.
7. Run proof only for ICORE public detail pages with the reviewed association adapter, then verify. DOAJ and imports are proof-ineligible.
8. Report freshness, access, and evidence gaps. Use incomplete analysis when material requested sources remain blocked or identity is unresolved.

Read these references as needed:

• references/source-policy.md before live lookup, refresh, or source addition.
• references/matching-policy.md for ambiguous identities or aliases.
• references/artifact-schema.md when inspecting or consuming run artifacts.
• references/proof-contract.md before capturing or validating proof.
• references/privacy-licensing-policy.md for authenticated, subscription, or restricted providers.
• references/deferred-roadmap.md for Phase D public-live and licensed-API work.

Runtime

POSIX:

bash "${AAS_RUNTIME_ROOT:-$HOME/.local/share/ai-agents-skills/runtime}/run_skill.sh" \
  skills/venue-ranking-evidence/run_venue_ranking_evidence.sh \
  lookup --dir /path/to/run --query "ISAAC" --source icore --offline

Windows PowerShell:

$runtime = if ($env:AAS_RUNTIME_ROOT) { $env:AAS_RUNTIME_ROOT } else { "$env:LOCALAPPDATA\ai-agents-skills\runtime" }
& "$runtime\run_skill.ps1" `
  "skills/venue-ranking-evidence/run_venue_ranking_evidence.ps1" `
  lookup --dir "$env:TEMP\venue-ranking-run" --query "TCS" --offline

On Windows, use the PowerShell command above; CMD runtime entrypoints are not published.

The POSIX wrapper honors VENUE_RANKING_EVIDENCE_PYTHON, then AAS_RUNTIME_PYTHON, before falling back to python3 and python. Browser proof capture requires Chromium, Chrome, or Edge. Proof marker verification also requires Poppler's pdftotext on PATH. Run doctor to inspect both.

Useful verbs:

• doctor
• sources list|show|check
• sources check --source <id> --data-file <id>=<path> (import preflight)
• sources validate|add --descriptor <file> --registry-dir <dir>
• lookup --dir <run> --query <text> [--source <id> ...]
• lookup ... --data-file <source-id>=<csv-or-json>
• lookup ... --venue-type journal|conference|... (repeatable)
• lookup ... --max-candidates N / --include-all-candidates
• proof --dir <run> --observation-id <id> [--venue-id <id>]
• report|verify|purge --dir <run>
• cache status|refresh|purge
• smoke

Live operations require both --allow-network and --allow-source <id>.

Offline ICORE cache demotes formerly-current rows to currentness-unconfirmed.

Output contract

lookup prints and writes delivery.json with:

• status: ready only when identity is resolved (match_status=matched), every requested source is satisfied (source_coverage_status complete or not-requested), and analysis is not incomplete.
• match_status: matched | ambiguous | unmatched
• source_coverage_status: complete | partial | empty | not-requested
• incomplete_analysis: true when coverage is partial/empty, identity is ambiguous, or a journal-path gap applies
• requested_sources / satisfied_sources / missing_or_blocked
• total_candidates / displayed_candidates / resolved_venue_id / ambiguity_requires_selection

For each candidate, show title, type, identifiers, aliases, match method, and one observation row per source assertion (kind, scheme, category/collection, value, edition/year, freshness, official URL). Index membership is not a rank.

Proof rules

Browser Print-to-PDF is not a pixel-identical screen capture. Preserve raw PDF, full-page PNG, sidecars, and manifest fields from runtime. Only verify can return VERIFIED. No login, CAPTCHA, or browser-profile reuse. Proof is limited to the public unauthenticated ICORE detail page.

Source extension boundary

User-added declarative CSV/JSON sources need validated descriptors without executable hooks or reviewed proof association adapters. New live adapters need reviewed built-in code and fixtures.

Built-in user-export normalized columns: required canonical_title and value; optional identity/metric fields as documented in references/artifact-schema.md. DOAJ raw public CSV is accepted via live fetch or --data-file doaj=... through the built-in DOAJ parser.