Loading consolidated registers and recipes...
Error loading progress database. Ensure that docs/dashboard_data.json is compiled and generated correctly.
Family Title
Product typeFamily definition details.
Dominant Limiting Factor
Limiting factor description.
Grouping Rationale
Rationale text.
▼ T4 Suitability Variable Mappings
Click to collapse/expand
Dimension:
| Variable | Dimension | Default Weight | Standard Dataset | Response Function / Params | Evidence Reconciliation & Justification | Action |
|---|
▼ Verbatim Evidence Claims
Click header to collapse/expand
▼ VONT Variable Ontology Register
Click to collapse/expand
Group:
| Variable ID | Label | Group | Canonical Unit | Min meaningful resolution | Sensitivity |
|---|
▼ T1 Spatial Dataset Registry
Click to collapse/expand
Access:
| Dataset ID | Dataset Name | Coverage | Format | Access Type / asset ID | Limitations |
|---|
Evidence & Methodology Network Diagram Force-directed relational network showing how literature sources translate to biophysical variables, suitability families, and target recipe tables.
Filter Settings
Visible Graph Entities:
Lineage Explorer
Hover over any node to highlight its connected neighborhood. Drag nodes to explore relationships. Use mouse scroll to Zoom and Pan. Double-click to pin/unpin nodes.
💡 Drag nodes to explore. Scroll to Zoom. Double-click any node to pin/unpin it.
Node Inspector
Hover or select a node in the network to inspect its metadata, details, and verbatim literature claims.
⚐ QA / QC — VERIFY-FLAG review Every EV row flagged by the relationship-verify pass — work through these before any threshold is trusted
ℹ How to review
AI INCLUDED — AI accepted it; spot-check it's faithful (AI-passed view)
MISMATCH — AI-flagged: a value contradicted the quote (auto-corrected)
UNSUPPORTED — AI-flagged: a value the quote never contained (stripped)
To review: set your handle → choose ok (keep) or drop (remove) + a reason → Apply & submit to main writes it to the register, re-gates, and opens a PR that auto-merges on green CI.
No live server? Use Download decisions CSV →
Per card: ok = keep · drop = remove (needs a coded reason) · ⚑ query = keep tentatively but raise an open question (write it in the note).
review.py apply.
Per card: ok = keep · drop = remove (needs a coded reason) · ⚑ query = keep tentatively but raise an open question (write it in the note).
Keys:
j/↓ next · ↑ prev · k ok · d drop · q query (ignored while you're typing in a field)Reason codes — what each means
smuggled_number — number not in this quote
table_error — mangled / mis-stitched table (cells merged across rows, garbled values)
wrong_variable — mis-labelled variable (SOM≠SOC)
wrong_practice — wrong NbS/practice (reforestation, pure forestry, carbon-only)
species_envelope — per-species range, not a practice rule (retag species-specific + keep, don't drop)
off_scope — not a suitability claim (site descriptors, wrong crop)
wrong_table — soft enabling-env → T3/M2b/next-steps, not T4
constrained_aoi — source only covers part of the land (e.g. “marginal land”) → understates where the NbS can go; don't use to set extent
quote_too_narrow — too little context
relationship_missed — suitability curve/directionality in source but not captured
insufficient_context — no caption/context or screengrab failed → cannot review
unusable_value — variable named but no usable rule: weight with no scale (“5 out of what?”) or land cover with no class list
speculative_evidence — speculation/assumption, not first/second-hand
false_flag — verifier wrong → keep (ok)
accepted_correction — fix accepted → ok
confirmed_pass — AI pass was correct → ok (passed mode)
missed_error — AI missed it → drop (passed mode)
other — explain in the note
Filters
ⓘ how the filters work
The Filter bar (right) has three orthogonal toggles, each with a live count — combine them for any permutation:
Origin (Any / ⚑ AI-flagged / ✓ AI-passed) · Reviewed by me (Any / Not / Yes) · Reviewed by others (Any / Not / Yes).
Common views: To review = me Not · 2nd opinion = me Not + others Yes · your own calls = me Yes, then the My call sub-filter (✓ Kept / ⚑ Queried / 🗑 Dropped).
Everything also follows the SCOPE below (Practice · Sub-practice · Evidence type · channel · access) — change any and all counts update together.
⚠ Gaps / 📊 Stats — buttons top-right.
Scope
Filter
📚 Stocktake & Tools
Catalogues of what's out there: Peer-reviewed = 220 sources from the pre-repo OpenAlex search (C/I/D benchmark); Tools / methods = the in-repo TOOL_tool_registry (codebases · platforms · method frameworks). In-repo paper-first / grey sweeps live under Discovery & Search Log.
Evidence & Discovery Register coverage, discovery progress, the registered sources behind it, and the PRISMA-lite search protocols that found them. Everything below follows the scope chips.
Practice (NbS)
Table / Step
Sub-Practice
0
Total Sources Logged
0
Swept / Ingested Sources
0
Total Evidence Claims
0
Compiled Recipes
Evidence Quality Signal Live counts from the Evidence Register — full review in the QA / QC Flags tab ▶ Open QA / QC Flags tab
TOTAL EV UNITS
—
VERIFY-FLAGS
—
UNREVIEWED
—
LOW CONFIDENCE
—
BY USE ROLE
—
▼ Coverage Matrix — Discovery, Evidence & Methodology Progress
Click to collapse/expand
Status:
Practice & sub-practice follow the scope chips above.
ℹ️ What this matrix shows & how to use it — click to expand
One group row per NbS (with its register roll-ups), one row per suitability family. Stock · Lit · Grey · Tools are the four discovery channels feeding T4 suitability; Claims and T4 rows are what reached the registers; Status is the family's methodology state.
Count badge
12found — candidate sources screened-in (SRCH n_included) awaiting extraction. Click → that cell's evidence units.
Stage pip
furthest stage reached, in order Searched → Extracted → Verified → Reviewed. Green = direct (family-targeted search). slate · gen = reached only by the practice-wide generic parent search. Click → that cell's search protocol.
Progress bar
green fill = share of the cell's found sources already extracted into the register.
extracted / found
| Suitability Family ▲ | Stock | Lit | Grey | Tools | Claims | T4 rows | Status |
|---|
Search protocol
Registered Sources
Filter Category:
Stage S logged in a discovery log · E registered in the source register · Xn n evidence claims extracted · ⚑ claims flagged for review · R human-reviewed. Filled = done, hollow = not yet.
Tier = source credibility (High / Medium / Low) from the screening rubric (evidence strength · transparency · venue · context · recency).
Rows come from the Source Register (SRC) scoped by the chips above — not only from a parsed discovery log — so every practice's sources appear. Click a row for its evidence claims + raw register JSON.
Tier = source credibility (High / Medium / Low) from the screening rubric (evidence strength · transparency · venue · context · recency).
Rows come from the Source Register (SRC) scoped by the chips above — not only from a parsed discovery log — so every practice's sources appear. Click a row for its evidence claims + raw register JSON.
| Source | Tier | Category | Geography / Context | Stages | Variables Extracted |
|---|
Evidence Claims for Source:
| Evidence ID | Variable | Relationship / Rules | Context | Verbatim Quote / Assertion | Page | Evidence Type |
|---|
Source Metadata JSON
▼ Search Protocol (PRISMA-lite) & Triage Decisions
Pick a practice (and optionally a sub-practice) in the scope chips above to load its PRISMA-lite discovery funnel, search queries and triage decisions.
PRISMA-lite Discovery Funnel
Queries & Databases Screened
Borderline & Triage Decisions (Human-in-the-Loop Review)
| Decision | Category / Source | Ambiguity / Reasoning | Proposal & Context Boundary Alignment |
|---|
▼ Evidence Sourcing & Translation Flow
Click to expand
Filter Target Tables:
Hover a node to trace its lineage. The source-level grid now lives in Registered Sources above.
▼ Register Analytics
Click to expandSources Extraction Status per NbS
Swept / Extracted
Pending / Sweeps remaining
Evidence Benchmark Tiers
100%
Traceable
Represented Study Countries / Contexts
Literature Sourcing Methods
Last compiled: 2026-06-12
•
Git commit:
abcdefg