Research libraryThe Industrial Data Spine
Chapter 14 of 14 · Part IV · Draft web edition
GFT should scale the spine only as ownership, meaning, recovery, reproducibility and net value are demonstrated in production.
The cloth is woven from this chapter's 4 exhibits: each weft is one exhibit's own data, normalised; each warp group is a section. The probe reads the exhibit's own value where it crosses. Procedural visualisation, not a measurement.
Read-out
Illustrative 0 to 18 month roadmap aligned with GFT's Cameroon textile programme phases. Entry and exit gates derive from S02, S04, S15, S16, S26, S36 and S44; task durations are planning assumptions, not measured factory benchmarks.
Unit: months from approval (illustrative planning window). Status: illustrative. Sources: S02, S04, S15, S16, S26, S36, S44. Illustrative figures are teaching examples, not measured factory results.
A small or mid-sized factory does not need a standalone data department to govern its operational records or run machine-learning models safely in production. Instead, it needs a staged, evidence-gated sequence that embeds data ownership, explicit event semantics, quality quarantine, tested offline recovery, reproducible model release and net value measurement directly into existing factory roles and management routines.[S02][S16][S36] In GFT's Cameroon textile programme, prepared on 31 January 2024 with source-refresh annotations retained in 2026, the sector context is defined by an acute local-processing gap: the inherited evidence register records that about 95 percent of Cameroon's cotton fibre is exported raw while about 5 percent is processed locally, against a national industrialisation target of 50 percent local processing by 2030 under the National Development Strategy.[R06][R07] Those two inherited entries were not refreshed in 2026 and their original publication dates were not recorded in the register, so they frame the strategic imperative rather than certifying a current factory baseline.[R06][R07] Closing that industrial gap requires factories whose daily production, quality, maintenance, order and resource records can be trusted from the first shift.
This closing chapter translates the preceding governance, architecture and MLOps controls into an illustrative 18-month implementation roadmap structured around six sequential stages and nine hard decision gates. Every duration in this chapter is an illustrative planning assumption designed to show sequencing and dependency, not an observed benchmark or fixed calendar promise.[S16][S36] Advancement from one stage to the next is governed by inspectable operational evidence rather than elapsed calendar time. When a gate fails because capture effort is excessive, quality rules are bypassed, backups cannot be restored, or a model candidate fails to beat a simple comparator net of lifecycle cost, the required governance action is to hold, redesign, postpone or stop the expansion.[S26][S44]
14.1
Read-out
Illustrative mapping of the six functional governance roles and three decision anchors across technical execution closeness (x-axis) and operational decision authority (y-axis), derived from role assignment in S02, human threshold setting in S11, AI risk integration in S14 and enterprise risk governance in S23. Coordinates are illustrative design positions for a small factory, not empirical measurements.
Unit: illustrative governance coordinates (0 to 10 scale). Status: illustrative. Sources: S02, S11, S14, S23. Illustrative figures are teaching examples, not measured factory results.
Read this exhibit in its chapter ↗14.1 · Introduction
The opening illustrative window from day 1 to day 30 establishes human accountability, bounds the initial scope to a handful of decision-critical records…
Exhibit 05, chapter 02: Domain owner
14.2 · Authority precedes software selection
Each domain owner signs a compact role charter that separates three distinct responsibilities: business accountability for meaning and access (the data owner ), recurring shop-floor verification and exception handling (the data steward ), and technical administration of spreadsheets, databases…
Exhibit 05, chapter 02: Domain owner
14.3 · Legal screening and baseline measurement start on day one
Because operational records touch workers and external partners, the first 30 days also establish the privacy and access boundary.
Exhibit 05, chapter 02: Model owner
14.1 · the bodyThe film above follows these sub-steps.
The opening illustrative window from day 1 to day 30 establishes human accountability, bounds the initial scope to a handful of decision-critical records, registers lawful processing purposes and freezes the baseline definitions against which every subsequent improvement claim will be judged.[S02][S36][S38] Small industrial teams frequently stall when they begin digitalisation by evaluating software platforms before deciding who owns a definition and which operational decision the record serves.[S02][S37] Survey evidence from the Organisation for Economic Co-operation and Development in the Western Balkans indicates that integrated operational tools such as enterprise resource planning and cloud systems are used by only 10 to 20 percent of surveyed small and medium enterprises, showing why a roadmap cannot assume integrated systems already exist (while noting that this survey covers the Western Balkans rather than Cameroon).[S34] Under ISO 8000-150, establishing appropriate roles and responsibilities is a foundational requirement for data quality management, and World Bank case evidence from small-firm adoption confirms that firms successfully adopting digital platforms assign an internal point person who champions adoption across functions.[S02][S37] During this illustrative first month, the factory director appoints one internal adoption champion with a named deputy and assigns one accountable domain owner for each of six core domains: production, quality, maintenance, orders, people and sustainability.[S02][S37]
Each domain owner signs a compact role charter that separates three distinct responsibilities: business accountability for meaning and access (the data owner), recurring shop-floor verification and exception handling (the data steward), and technical administration of spreadsheets, databases, brokers or backups (the data custodian).[S02][S23] In a constrained team, one person may hold more than one role across domains, provided conflicts of interest are declared and no technical custodian can silently alter a business formula or approve a high-consequence model release alone.[S02][S11][S14] Line stewards receive an explicit, illustrative weekly time allowance within their normal shift duties so that reviewing missing logs or quarantining malformed entries is scheduled operational work rather than unpaid overtime.[S02][S35]
Rather than cataloguing every machine tag or administrative spreadsheet in the enterprise, the champion and domain owners select an illustrative initial scope of five critical operational records governing daily shift performance: hourly line output, defect inspection events, unplanned machine stoppages, order and lot dispatch records, and shift attendance aggregates.[S05][S08][S38] For each selected record, the domain owner freezes a versioned entry in the KPI dictionary specifying purpose, formula, input elements, unit, time behaviour, grain and exclusions under ISO 22400-2.[S01][S38] Freezing these definitions before pipeline or model work begins prevents later disputes over whether rework counts as good output or whether planned breaks reduce availability.[S38]
Because operational records touch workers and external partners, the first 30 days also establish the privacy and access boundary.[S29][S31][S1401] Under Regulation (EU) 2016/679, personal data must be adequate, relevant and limited to what is necessary for the stated purpose, while World Wide Web Consortium best practices emphasise that security, commercial sensitivity and privacy govern what data and metadata may be shared.[S29][S1401] In Cameroon, Law No. 2024/017 of 23 December 2024 states in Section 19(1) that the processing of personal data shall be subject to prior authorisation by the Personal Data Protection Authority.[S31] Before collecting or repurposing any worker-linked telemetry, attendance record or video feed in this illustrative 30-day window, the factory records each processing purpose in a local register, strips unnecessary personal identifiers from operational aggregates, and obtains current local legal advice on statutory authorisation duties.[S29][S30][S31]
Finally, the team records a pre-intervention operating baseline across process performance and data handling, applying the World Bank principle that embedding monitoring and evaluation from the outset is essential for tracking impact and learning.[S36] Over an illustrative two-week to four-week observation window, stewards measure shift minutes spent entering and reconciling logs, capture-defect recurrence, correction turnaround time, backup restoration capability, and baseline levels for first-pass yield, line availability and order fulfilment.[S26][S36][S38] Without this frozen baseline, neither the manual data spine nor any subsequent model can prove falsifiable operational value.[S36][S43]
As shown in Exhibit 60, the six illustrative stages run across an 18-month planning window from month 0 to month 18, with each stage bounded by explicit entry and exit conditions.[S16][S36] The chart is not an automatic calendar escalator: if Stage 2 fails to prove offline buffering and backup restoration at month 3, Stage 3 and Stage 4 remain locked until the recovery defect is fixed and re-verified.[S26][S44] What Exhibit 60 cannot establish is the exact pace of a specific site, because actual durations depend on shift complexity, staffing stability, power reliability and legal review timelines.[S35][S42]
14.4
Once authority, scope and baseline definitions pass Gate 1 at day 30, the illustrative second stage from day 31 to day 90 turns the factory's manual and semi-automated logs into a governed system of record.[S04][S08][S09] GFT's Cameroon textile programme starts with structured manual and spreadsheet-backed capture on the shop floor rather than waiting for full machine instrumentation. That approach succeeds only when manual capture enforces the same identifier discipline, event semantics, quality checks and continuity rehearsals that later automated feeds will inherit.[S04][S08]
scope and baseline definitions pass Gate 1 at day 30
S04, S08, S09 · illustrative
Between illustrative days 31 and 60, the domain owners and custodian issue versioned data contracts for the critical records selected in Stage 1.[S04][S10] Every record is anchored to immutable canonical identifiers for site, area, line, asset, order, lot, style, operation, defect code and shift, excluding personal worker identifiers unless a documented, legally reviewed purpose requires them.[S05][S08][S29] Following GS1 EPCIS Release 2.0, operational occurrences such as defect observations, machine faults, lot splits or carton dispatches are recorded as discrete events carrying event time, ingestion time, object identity, location and business step rather than overwriting mutable status cells.[S08] Provenance links follow the World Wide Web Consortium PROV-O and PROV-DM specifications so every record retains the entity, activity and agent involved in producing or modifying it.[S09][S1403]
In the same illustrative window, the factory activates capture and ingestion quality gates under ISO/IEC 5259-3 and ISO/IEC 25012 across five dimensions: validity, completeness, consistency, timeliness and uniqueness.[S03][S04] Missing readings are tagged explicitly as missing, engineering approximations as estimated, and supervisor amendments as manually_corrected with a superseding provenance link, never silently replaced with a numeric zero.[S04][S09] Failed records are routed into a quarantine table and assigned to the originating line steward for correction, preventing upstream data cascades from contaminating curated summaries.[S04][S19] Each recurring defect, its decision impact, its root cause and its verification date are tracked in a published quality register informed by the W3C Data Quality Vocabulary and iterative quality guidance.[S41][S1402]
Between illustrative days 61 and 90, before any machine-learning model is admitted, the team proves that the manual spine survives infrastructure disruption.[S26][S27][S33] Drawing on store-and-forward buffering patterns documented for offline edge devices, local capture sheets and edge queues buffer timestamped records durably during network outages and replay them idempotently without duplication when connectivity returns.[S07][S08][S33] Simultaneously, the custodian maintains offline encrypted backups of raw logs, master tables, KPI definitions and validation rules, and executes a live restoration drill into an isolated environment alongside an incident-response exercise as recommended by CISA and NIST SP 800-61 Revision 3.[S26][S27]
Read-out
| Days 1 to 10: Factory director | Six-role governance charter and champion mandate | Signed charter with deputy and steward time allowance | Weekly operations review |
|---|---|---|---|
| Days 11 to 20: Domain owners | Version 1.0 KPI dictionary for priority records | Frozen formulas, units, grains and exclusions | Weekly operations review |
| Days 15 to 25: Champion and counsel | Local personal-data processing register | Minimised fields and Cameroon Law 2024/017 screening | Executive risk sign-off |
| Days 20 to 30: Line stewards | Four-week operating and data-effort baseline | Logged capture minutes, error rates and baseline KPIs | Gate 1 review at day 30 |
| Days 31 to 50: Custodian and owners | Canonical ID registry and event data contracts | Schema check on production, defect and stop logs | Daily shift handover |
| Days 51 to 75: Line stewards | Active ingestion quarantine and quality register | Quarantined records resolved with PROV-O lineage | Weekly quality cadence |
| Days 76 to 90: Custodian and OT lead | Disconnection buffer test and backup restoration drill | Zero duplicate replay and verified isolated restore | Gate 2 review at day 90 |
Illustrative owner-level action list for days 0 to 30 and days 31 to 90, with deliverable, verification evidence and escalation route derived from S02, S04, S26, S36 and S38. Day windows are illustrative planning ranges.
Unit: illustrative 90-day action register. Status: illustrative. Sources: S02, S04, S26, S36, S38. Illustrative figures are teaching examples, not measured factory results.
The sequence in Exhibit 61 makes every deliverable in the first 90 days inspectable by naming the owner, artefact, verification proof and escalation forum.[S02][S36][S38] Rather than creating a second governance committee, each action routes exceptions into the factory's daily shift handover and weekly operations review (wp02).[S02][S23] Because Exhibit 61 is an illustrative template, a single-line plant can compress administrative steps, whereas a multi-site spinning, weaving and garment operation must validate identifiers across site boundaries before closing Gate 2.[S05][S08]
14.7
Related · chapter 055 paragraphs step it
Read-out
Derived minimum architecture separating capture, immutable raw evidence, contextual canonical records, published data products and operational decisions, based on interface boundaries in S05, provenance modelling in S09 and S0501, smart manufacturing standards in S10 and pipeline lineage in S15.
Unit: . Status: derived. Sources: S05, S09, S10, S15, S0501.
Read this exhibit in its chapter ↗Passing Gate 2 at day 90 proves that critical shop-floor events are captured under stable identifiers, checked by active quarantine rules and recoverable from offline backups.[S04][S26] The illustrative third stage from month 4 to month 6 converts those raw logs into reusable, governed data products so downstream consumers rely on one canonical core instead of fragile, application-specific extraction scripts.[S09][S10][S39] Workshop evidence from the National Institute of Standards and Technology notes that manufacturers have reported data-processing costs for a single application reaching as high as 70 percent of total cost.[S39] Treated carefully as an upper-bound warning from semiconductor manufacturing rather than a universal SME average, that 70 percent figure underscores the cost penalty of building separate pipelines for every report or model.[S18][S39]
During illustrative months 4 and 5, the custodian and domain owners separate three storage layers: immutable raw evidence, contextualised canonical records and published consumption products.[S05][S09][S15] Raw shift logs, inspection sheets and machine messages are stored append-only with ingestion timestamps and digests so original evidence is never overwritten.[S09][S1403] Above the raw layer, deterministic transformations join events to versioned master data (assets, styles, calendars and defect taxonomies) and apply quality rules to publish curated operational tables.[S04][S06][S08] When a steward corrects a quarantined entry, the correction appends a superseding event linked via prov:wasRevisionOf and recomputes the curated product with full lineage.[S09][S15]
From these curated tables, the spine publishes controlled consumption views without duplicating business formulas: production and downtime views for OEE and line balancing (wp02, wp15), defect and genealogy views for quality and traceability (wp03, wp05), and utility and material views for resource and carbon accounting (wp06, wp16).[S05][S08][S38][S41] Each published view carries Data Quality Vocabulary metadata disclosing completeness, freshness and estimation share so consumers never mistake an estimate for a metered reading.[S41][S1402]
Before Stage 3 closes at month 6, the factory completes three verification checks.[S25][S32][S38] First, over an illustrative four-week parallel run, the team reconciles KPI figures from the curated products against legacy spreadsheet calculations under the frozen ISO 22400-2 definitions.[S09][S38] Second, the security approver and custodian review access under NIST SP 800-53 Revision 5 least-privilege controls and CISA cybersecurity goals, confirming that operators, supervisors, analysts, vendors and automated services use separate, narrowly scoped credentials.[S24][S25][S1401] Third, the team executes a portability export test: applying EU Data Act principles where applicable and explicit contract clauses elsewhere, the custodian exports six months of raw events, curated tables, schemas, KPI definitions and PROV-O lineage into open formats and verifies that an independent script reproduces the monthly management report without vendor software.[S06][S08][S09][S32]
Interlude · the cloth half-woven
Woven: sections 14.0 to 14.7Still to weave: 14.10 to 14.20
Read-out
Inherited GFT programme evidence base prepared 31 January 2024. R06 and R07 were not refreshed in 2026 and their original publication dates are unrecorded. Figures describe sector-level baseline and policy ambition, not a current GFT factory measurement or causal productivity claim.
Unit: percent of cotton fibre. Status: sourced. Sources: R06, R07.
Read this exhibit in its chapter ↗14.10
G1: Role charter and KPI freeze
G1: Role charter and KPI freeze: Roadmap sequence position (1 early to 10 late) 1.2, Downstream blast radius if bypassed (1 to 10) 7.5. Quadrant: Foundational governance gates · derived
Only after Gate 3 confirms trusted, portable data products at month 6 does the roadmap admit a first machine-learning candidate in the illustrative window from month 7 to month 9.
G2: Event contracts and quarantine
G2: Event contracts and quarantine: Roadmap sequence position (1 early to 10 late) 2.4, Downstream blast radius if bypassed (1 to 10) 8. Quadrant: Foundational governance gates · derived
G4: Curated products and KPI parity
G4: Curated products and KPI parity: Roadmap sequence position (1 early to 10 late) 4.6, Downstream blast radius if bypassed (1 to 10) 7.2. Quadrant: Foundational governance gates · derived
G5: Access review and export test
G5: Access review and export test: Roadmap sequence position (1 early to 10 late) 5.5, Downstream blast radius if bypassed (1 to 10) 6.8. Quadrant: Release and scale gates · derived
G6: Model admission and snapshot
G6: Model admission and snapshot: Roadmap sequence position (1 early to 10 late) 6.6, Downstream blast radius if bypassed (1 to 10) 7.8. Quadrant: Release and scale gates · derived
G8: Alert runbook and net value audit
G8: Alert runbook and net value audit: Roadmap sequence position (1 early to 10 late) 8.6, Downstream blast radius if bypassed (1 to 10) 8.5. Quadrant: Release and scale gates · derived
G9: Second-use reuse and exit drill
G9: Second-use reuse and exit drill: Roadmap sequence position (1 early to 10 late) 9.5, Downstream blast radius if bypassed (1 to 10) 7.4. Quadrant: Release and scale gates · derived
Derived gate register placing nine programme gates across sequence position (x-axis, 1 to 10) and downstream decision and safety blast radius if bypassed (y-axis, 1 to 10), synthesised from S02, S04, S09, S11, S15, S17, S26, S32, S36 and S44.
Unit: derived ordinal scale from 1 (initial) to 10 (factory-wide). Status: derived. Sources: S02, S04, S09, S11, S15, S17, S26, S32, S36, S44.
14.10 · the bodyExhibit 62 above; its figures are illustrative where marked.
Only after Gate 3 confirms trusted, portable data products at month 6 does the roadmap admit a first machine-learning candidate in the illustrative window from month 7 to month 9.[S04][S16][S43] Industrial MLOps research shows that starting infrastructure and target automation levels drive implementation complexity, making gradual capability growth essential for small teams.[S16][S42] Following GFT's Cameroon textile programme sequence, the recommended first use case is a bounded shop-floor problem with rapid ground-truth feedback, such as computer-vision defect inspection (wp03), whereas history-dependent demand forecasting (wp04) waits until clean multi-season order history accumulates.[R29][S04][S43]
Month 7 opens with the model admission screen grounded in the six-phase CRISP-DM lifecycle and explicit SME project-postponement criteria.[S43][S44] The operating owner specifies the target decision, acting role, response window, operational technology safety boundary under NIST SP 800-82 Revision 3, and worker-protection rules under the International Labour Office code of practice.[S22][S30][S43] The team freezes the pre-model baseline and defines a simple, transparent comparator (such as a statistical rule or standard inspection checklist) that the candidate must beat net of lifecycle cost.[S36][S38][S43] Builder and custodian then record the candidate in a lightweight reproducibility stack comprising version control, immutable dataset snapshots with ISO/IEC 5259-3 quality manifests, an experiment execution log, and a model registry binding data, code, vendor version and evaluation metrics to one candidate identifier.[S04][S13][S15][S20]
During illustrative months 8 and 9, the candidate runs exclusively in silent shadow mode on live inputs, logging timestamped predictions without displaying outputs to operators or altering shop-floor actions.[S15][S16][S22] Each week, the builder and steward evaluate shadow predictions against verified ground truth across local slices (fabric shade, style, shift, lighting and missing-data state) so aggregate scores cannot hide line-specific failures.[S04][S11][S19] Before Gate 7 closes at month 9, the team executes a live rollback and manual-fallback drill, proving that model suspension and reversion to manual operation complete cleanly within the target recovery window.[S15][S22][S27] Operating and risk owners then judge shadow slice results, comparator margin, support cost and rollback evidence against human-set thresholds, postponing or stopping any candidate that fails.[S11][S14][S44]
The derived gate register in Exhibit 62 maps the nine decision gates (G1 through G9) across sequence position (horizontal axis, 1 to 10) and downstream operational and safety blast radius if bypassed (vertical axis, 1 to 10).[S02][S11][S26][S44] Early foundational gates (G1 to G3) score high on blast radius because unowned definitions or unrecoverable records corrupt every subsequent report and model, while G7 carries the highest score (9.2 on the derived 10-point scale) because it guards the transition from silent shadow testing to live shop-floor interaction.[S04][S11][S19][S26] Coordinates in Exhibit 62 are analytical ordering indices rather than empirical measurements; any control-connected use case requires even stricter scrutiny under ISA/IEC 62443 and NIST SP 800-82 Revision 3.[S22][S28]
14.13
Clearing Gate 7 at month 9 opens the illustrative fifth stage from month 10 to month 12, moving the candidate from shadow mode into controlled, operator-assisted production.[S12][S13][S16] Initial release remains advisory: the model flags a potential defect or maintenance anomaly, while a trained operator or supervisor retains explicit authority and time to inspect, accept or override the output.[S11][S22][S30] Where outputs relate to worker activity, the International Labour Office principle that employment decisions must not rest solely on automated personal-data processing is enforced as a hard rule with logged human review.[S30]
Month 10 begins by issuing an ISO/IEC 42001 and ISO/IEC 5338 release package documenting approved model and dataset versions, intended use, slice limits, operator instructions, fallback procedures and review dates, paired with task-based workstation training anchored in tangible daily work.[S12][S13][S35][S40] The custodian activates threshold monitors across service health, feature and prediction drift, and delayed label accuracy.[S04][S17][S21] Following Azure ML monitoring architecture and concept-drift research, crossing a statistical threshold triggers a rehearsed diagnostic runbook rather than automatic retraining: the assigned role verifies whether the shift reflects an ingestion fault, a product-mix change or genuine concept drift, and either repairs the feed, invokes manual fallback or opens a governed retraining candidate.[S13][S17][S21][S27]
At month 12 (Gate 8), the champion, operating owner and controller audit three-month production results against the Stage 1 baseline across adoption behaviour (timely capture, completeness, correction recurrence, meeting use and override quality), operational KPIs and net financial return after subtracting ongoing preparation, review, integration and support costs.[S18][S36][S38][S39]
Illustrative net-value gate calculation for a single-line defect-inspection assistant at month 12. Consider an illustrative garment plant evaluating an operator-assisted vision station at Gate 8 after three months of operation (months 10 to 12). All figures are explicitly illustrative assumptions demonstrating the accounting method, not observed GFT factory results or forecasts.
14.16
When Gate 8 confirms positive net value at month 12, the illustrative final stage from month 13 to month 18 proves the economic core of the spine: reusing governed identifiers, curated events and lineage across additional decisions without rebuilding capture pipelines.[S08][S09][S10][S39] Stage 6 prevents post-pilot pipeline sprawl by requiring canonical reuse before any new analytics tool is approved.[S18][S19]
During illustrative months 13 to 15, the team connects a second decision to existing curated tables without adding shop-floor forms: canonical order, cotton-lot and energy-meter events feed product carbon footprints (wp16) and provenance certificates (wp05), while governed downtime and quality-state events feed line balancing and maintenance (wp02, wp07, wp15).[S05][S08][S38][S41] Simultaneously, MLOps automation is added selectively where Stage 4 and Stage 5 logs show recurring manual toil, such as automating weekly slice-evaluation scripts while keeping retraining and promotion strictly human-approved under NIST AI RMF 1.0 and ISO/IEC 23894.[S11][S14][S16][S42]
Months 15 to 18 also test history-dependent use cases such as demand and inventory forecasting (wp04).[S04][S43] Although McKinsey's 2025 fashion research cites inventory-cost reductions of 35 to 40 percent and forecast-error cuts of 20 to 50 percent from AI at scale, a Cameroon factory cannot capture those gains until it accumulates clean multi-cycle order and stockout records.[R27][S04][S19] Under GFT's roadmap, a forecaster enters Stage 6 shadow testing only if Gate 6 verifies at least 12 to 15 months of governed order and stockout history and the candidate beats a simple moving-average or seasonal comparator.[S04][S38][S43][S44] Before Gate 9 closes at month 18, the plant completes an ISO/IEC 42001 review, retires obsolete artefacts, and migrates one curated dataset and model candidate to an alternate environment using only archived contracts, PROV-O lineage and open exports.[S06][S09][S12][S15][S32]
14.19
Read-out · derived
Derived monthly measurement pack synthesised from S02, S04, S15, S17, S26, S36 and S1402. Tile 2 pairs semantic contract coverage (Measure 2) and quality-gate resolution (Measure 3) so all seven capabilities are surfaced without universal numeric targets before site baseline.
Unit: derived monthly governance and MLOps capability pack. Status: derived. Sources: S02, S04, S15, S17, S26, S36, S1402.
missing or estimated share under ISO/IEC 5259-3 and the W3C Data Quality Vocabulary.[S04][S41][S1402]The six tiles in Exhibit 63 surface all seven monthly capability measures (combining Measures 2 and 3 in Tile 2 for legibility) without prescribing arbitrary universal percentages before a site's Stage 1 baseline is recorded.[S04][S36][S1402] Imposing unmeasured targets on day one incentivises fabricated entries; publishing actual baselines, exception counts and estimation shares builds durable trust.[S19][S41][S1402]
14.20
Across all 14 chapters of wp14, one principle governs: a small factory captures durable value from data and machine learning only by managing the path from physical event to decision as one auditable production spine.[S05][S09][S15][S43] That recommendation distils into five rules of disciplined portability:
By governing records once, wp14 supplies the evidentiary foundation across the GFT white-paper library and tools: daily management (wp02, kpi-cascade, oee-line), automated defect inspection (wp03, inspection-economics), cold-start demand forecasting (wp04, cold-start-forecast), lot traceability (wp05), utility efficiency (wp06), Scope 1, 2 and 3 carbon accounting (wp16, footprint), maintenance reliability (wp07), workforce capability (wp09), laboratory workflows (wp12), industrial engineering (wp15), compliance (wp17) and the edge-to-cloud topology (wp18, edge-architecture, build-or-buy).[S05][S08][S09][S38][S41] A small factory earns the right to scale artificial intelligence not by building a data bureaucracy, but by capturing the truth of its work on the first shift, governing every change of meaning, proving recovery under failure, and keeping every model accountable to the people who run the line.[S02][S11][S26][S36]
Check the record
Every knot is a source this chapter cites; select one to read its record, or filter the register by grade. Publication dates and inherited-source limits stay explicit. Read the paper's full evidence summary.
52 sources · 373 citation marks on this pageMost cited: S04, 22 times
Grade AGrade BGrade CKnot size: citations on this page
All 52 records shown
R06 · Grade C · inherited record
Inherited evidence R06
About 95 percent of cotton fibre exported raw against 5 percent processed locally; industrialisation target of 50 percent local processing.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R07 · Grade C · inherited record
Inherited evidence R07
SND30 restates the 50 percent local-processing target by 2030.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R12 · Grade C · inherited record
Inherited evidence R12
CICAM capacity utilisation, July 2023 report: Garoua 33.7 percent, CICAM 1 48.9 percent, CICAM 2 10.83 percent; underutilisation cost about CFAF 2 billion in 2020.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R27 · Grade B · inherited record
Inherited evidence R27
AI at scale in fashion: inventory cost down 35 to 40 percent; forecast accuracy up 20 to 30 percent; revenue per customer up 15 to 25 percent; forecast errors cut 20 to 50 percent; lost sales down up to 65 percent; 75 percent of executives prioritising; sector operating-profit potential US$150 to 275 billion over three to five years.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R29 · Grade C · inherited record
Inherited evidence R29
Manual defect inspection 60 to 70 percent accurate, missing 20 to 30 percent of defects; computer vision above 90 percent across more than 40 defect types; line-scan inspection up to 60 metres a minute.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
c14-S01 · Grade A
ISO 8000-1:2022: Data quality, Part 1: Overview
The series defines which characteristics of data are relevant to data quality, specifies requirements applicable to those characteristics, and provides guidelines for improving data quality.
2022. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S02 · Grade A
ISO 8000-150:2022: Data quality, Part 150: Data quality management: Roles and responsibilities
This document specifies the key considerations for organizations that are establishing appropriate roles and responsibilities for data quality management.
2022. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S03 · Grade A
ISO/IEC 25012:2008: Software engineering, Software product Quality Requirements and Evaluation, Data quality model
Data quality requirements and data quality measures can be categorized with the data quality characteristics in 5.2 and used by an evaluation process to analyse data independently from other computer system's components.
2008. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S04 · Grade A
ISO/IEC 5259-3:2024: Artificial intelligence, Data quality for analytics and machine learning, Part 3: Data quality management requirements and guidelines
ISO/IEC 5259-3 defines the requirements and guidance for managing the quality of data used in analytics and machine learning (ML).
2024. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S05 · Grade A
IEC 62264-1:2013, Edition 2.0: Enterprise-control system integration, Part 1: Models and terminology
This part of IEC 62264 provides standard models and terminology for describing the interfaces between the business systems of an enterprise and its manufacturing operations and control systems.
2013. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S06 · Grade A
OPC Unified Architecture, Part 1: Overview and Concepts, OPC 10000-1, version 1.05.06
It provides a comprehensive framework for modelling information, secure communication, and ensuring interoperability between diverse systems and devices.
2025-10-22. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S07 · Grade A
MQTT Version 5.0
MQTT is a Client Server publish/subscribe messaging transport protocol.
2019-03-07. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S08 · Grade A
EPCIS Standard, Release 2.0
The goal of EPCIS is to enable disparate applications to create and share visibility event data, both within and across enterprises.
2022. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S09 · Grade A
PROV-O: The PROV Ontology
It provides a set of classes, properties, and restrictions that can be used to represent and interchange provenance information generated in different systems and under different contexts.
2013-04-30. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S10 · Grade A
Current Standards Landscape for Smart Manufacturing Systems, NISTIR 8107
Standards are fundamental 1) to facilitate the delivery of the right information at the right time, 2) to enable actions based on that information and 3) to reduce risk of technology adoption and development.
2016-02-23. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S11 · Grade A
Artificial Intelligence Risk Management Framework, AI RMF 1.0, NIST AI 100-1
Human judgment should be employed when deciding on the specific metrics related to AI trustworthiness characteristics and the precise threshold values for those metrics.
2023-01-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S12 · Grade A
ISO/IEC 42001:2023: Information technology, Artificial intelligence, Management system
ISO/IEC 42001 is an international standard that specifies requirements for establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System (AIMS) within organizations.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S13 · Grade A
ISO/IEC 5338:2023: Information technology, Artificial intelligence, AI system life cycle processes
This document defines a set of processes and associated concepts for describing the life cycle of AI systems based on machine learning and heuristic systems.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S14 · Grade A
ISO/IEC 23894:2023: Information technology, Artificial intelligence, Guidance on risk management
The guidance also aims to assist organizations to integrate risk management into their AI-related activities and functions.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S15 · Grade B
MLOps: Continuous delivery and automation pipelines in machine learning
Information about each execution of the ML pipeline is recorded in order to help with data and artifacts lineage, reproducibility, and comparisons.
2024-08-28. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S16 · Grade B
MLOps Maturity Model
Use this model to develop your MLOps capability gradually instead of facing the full complexity of mature implementation upfront.
2026-06-03. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S17 · Grade B
Model monitoring in production
You can set thresholds for these metrics to trigger alerts about model or data anomalies via Azure Machine Learning or Azure Event Grid.
2026-01-27. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S18 · Grade B
Hidden Technical Debt in Machine Learning Systems
Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems.
2015. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S19 · Grade B
Everyone wants to do the model work, not the data work: Data Cascades in High-Stakes AI
Data cascades are pervasive (92% prevalence), invisible, delayed, but often avoidable.
2021. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S20 · Grade B
Machine Learning Operations: Overview, Definition, and Architecture
MLOps includes several aspects, such as best practices, sets of concepts, and development culture.
2022-05-14. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S21 · Grade B
Learning under Concept Drift: A Review
Concept drift describes unforeseeable changes in the underlying distribution of streaming data over time.
2020-04-13. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S22 · Grade A
Guide to Operational Technology Security, NIST SP 800-82 Revision 3
This document provides guidance on how to secure operational technology (OT) while addressing their unique performance, reliability, and safety requirements.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S23 · Grade A
The NIST Cybersecurity Framework 2.0
Governance activities are critical for incorporating cybersecurity into an organization's broader enterprise risk management (ERM) strategy.
2024-02-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S24 · Grade A
Cross-Sector Cybersecurity Performance Goals, Version 2.0
They aim to help critical infrastructure entities, particularly small and medium organizations, get started on their path toward a strong cybersecurity posture.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S25 · Grade A
Security and Privacy Controls for Information Systems and Organizations, NIST SP 800-53 Revision 5
Employ the principle of least privilege, allowing only authorized accesses for users (or processes acting on behalf of users) that are necessary to accomplish assigned organizational tasks.
2020-12-10. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S26 · Grade A
#StopRansomware Guide
Maintain offline, encrypted backups of critical data, and regularly test the availability and integrity of backups in a disaster recovery scenario [CPG 2.R].
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S27 · Grade A
Incident Response Recommendations and Considerations for Cybersecurity Risk Management: A CSF 2.0 Community Profile, NIST SP 800-61 Revision 3
Procedures can be tested or exercised periodically to verify their accuracy and can be used to help train new personnel.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S28 · Grade A
ISA/IEC 62443 Series of Standards
The ISA/IEC 62443 series addresses the security of industrial automation and control systems (IACS) throughout their lifecycle (which applies to all automation and control systems, not only industrial).
2026. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S29 · Grade A
Regulation (EU) 2016/679, General Data Protection Regulation
The personal data should be adequate, relevant and limited to what is necessary for the purposes for which they are processed.
2016-04-27. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S30 · Grade A
Protection of workers' personal data: An ILO code of practice
Decisions concerning a worker should not be based solely on the automated processing of that worker's personal data.
1997. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S31 · Grade A
Law No. 2024/017 of 23 December 2024 relating to personal data protection in Cameroon
The processing of personal data shall be subject to prior authorization by the Personal Data Protection Authority.
2024-12-23. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S32 · Grade A
Regulation (EU) 2023/2854 on harmonised rules on fair access to and use of data, Data Act
This Regulation ensures that users of a connected product or related service in the Union can access, in a timely manner, the data generated by the use of that connected product or related service and that those users can use the data, including by sharing them with third parties of their choice.
2023-12-13. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S33 · Grade B
Understand extended offline capabilities for IoT Edge devices, modules, and child devices
If your devices go offline, the IoT Edge parent device stores all device-to-cloud messages until the connection is reestablished.
2024-06-10. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S34 · Grade B
Uptake of digital tools among SMEs: Western Balkans Enterprise Survey
Tools that support more integrated business operations, such as cloud computing, enterprise resource planning (ERP) and customer relationship management (CRM) systems, are used by between 10-20% of surveyed firms, while e-commerce and supplier-customer management systems are less common.
2026-06-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S35 · Grade B
Employee empowerment through digital tools: The case of a German bakery SME
However, following managing director Holste the change of traditional processes into digital ones requires sizeable efforts that do not directly pay off but involve a longer learning process.
2022-11-21. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S36 · Grade B
The Missing Element of Firm Digitalization: Lessons for the EU Member States: Best practices for enhancing MSMEs Digitalization and Managerial Capacity Building
Embedding monitoring and evaluation throughout each step of the program design and implementation process is crucial for generating knowledge, tracking impact, and improving learning for the further improvement of interventions pursuing the digitalization of MSMEs.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S37 · Grade B
Challenges with digital technology adoption by SMEs: A case study from Viet Nam
Firms that successfully took up the platform assigned a point person who championed adoption within the firm.
2025-05-20. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S38 · Grade A
ISO 22400-2:2014: Automation systems and integration, Key performance indicators for manufacturing operations management, Part 2: Definitions and descriptions
The KPIs are presented by means of their formula and corresponding elements, their time behaviour, their unit/dimension and other characteristics.
2014. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S39 · Grade A
Artificial Intelligence with Open and Scaled Data Sharing in Semiconductor Manufacturing: Workshop Report
Also, manufacturers have reported that the cost of data processing for a single application can be as high as 70% of the total cost.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S40 · Grade B
The Digitization of TVET and Skills Systems
We suggest that digital transformation strategies have a better chance of success if they are developed to support tangible pilots that can demonstrate the practical use of emerging technologies with a short timeframe (two to three years maximum) and inbuilt mechanisms to deploy technology that provides ongoing data on what works and what does not to the sponsoring stakeholders.
2020. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S41 · Grade B
Technical Guidance for Calculating Scope 3 Emissions, version 1.0
Collecting data, assessing data quality, and improving data quality is an iterative process.
2013. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S42 · Grade B
Is MLOps different in Industry 4.0? General and Specific Challenges
Also the starting-level of the technical infrastructure as well as the aimed at degree of automation influence very significantly the overall complexity of implementing MLOps in Industry 4.0.
2022. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S43 · Grade B
Ensuring the Robustness and Reliability of Data-Driven Knowledge Discovery Models in Production and Manufacturing
Briefly, CRISP-DM divides the data mining-related knowledge discovery process into six phases: 1) business understanding, 2) data understanding, 3) data preparation, 4) modeling, 5) evaluation and 6) deployment.
2021-06-14. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S44 · Grade B
Supporting SME companies in mapping out AI potential: a Finnish AI development case
One must remember that in IT investments, the decision to postpone, e.g. an AI project, is also valid if resources, skills, costs, or expected benefits are not met.
2024-07-27. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S1401 · Grade A
Data on the Web Best Practices
Not all data and metadata should be shared openly, however. Security, commercial sensitivity and, above all, individuals' privacy need to be taken into account.
2017-01-31. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S1402 · Grade A
Data on the Web Best Practices: Data Quality Vocabulary
This document provides a framework in which the quality of a dataset can be described, whether by the dataset publisher or by a broader community of users.
2016-12-15. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c14-S1403 · Grade A
PROV-DM: The PROV Data Model
Provenance is information about entities, activities, and people involved in producing a piece of data or thing, which can be used to form assessments about its quality, reliability or trustworthiness.
2013-04-30. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗S02 International Organization for Standardization. 2022.
S16 Microsoft Learn, Azure Architecture Center. 2026-06-03.
S36 World Bank. 2023.
R06 Cameroon Trade Hub. Publication date unrecorded; inherited source-text check.
R07 Ecofin Agency. Publication date unrecorded; inherited source-text check.
S26 CISA, MS-ISAC, NSA and FBI. 2023.
S44 The Journal of Technology Transfer, Springer Nature. 2024-07-27.
S38 International Organization for Standardization. 2014.
S37 World Bank. 2025-05-20.
S34 OECD. 2026-06-26.
S23 National Institute of Standards and Technology. 2024-02-26.
S11 National Institute of Standards and Technology. 2023-01-26.
S14 ISO/IEC. 2023.
S35 OECD. 2022-11-21.
S05 International Electrotechnical Commission. 2013.
S08 GS1 AISBL. 2022.
S01 International Organization for Standardization. 2022.
S29 European Parliament and Council of the European Union. 2016-04-27.
S31 Presidency of the Republic of Cameroon. 2024-12-23.
S1401 World Wide Web Consortium. 2017-01-31.
S30 International Labour Office. 1997.
S43 Frontiers in Artificial Intelligence. 2021-06-14.
S04 ISO/IEC. 2024.
S15 Google Cloud. 2024-08-28.
S42 SCITEPRESS, IN4PL 2022. 2022.
S09 World Wide Web Consortium. 2013-04-30.
S10 National Institute of Standards and Technology. 2016-02-23.
S1403 World Wide Web Consortium. 2013-04-30.
S03 ISO/IEC. 2008.
S19 Association for Computing Machinery, CHI 2021. 2021.
S41 GHG Protocol, World Resources Institute and World Business Council for Sustainable Development. 2013.
S1402 World Wide Web Consortium. 2016-12-15.
S27 National Institute of Standards and Technology. 2025.
S33 Microsoft. 2024-06-10.
S07 OASIS Open. 2019-03-07.
S39 National Institute of Standards and Technology. 2025.
S18 Advances in Neural Information Processing Systems. 2015.
S06 OPC Foundation. 2025-10-22.
S25 National Institute of Standards and Technology. 2020-12-10.
S32 European Parliament and Council of the European Union. 2023-12-13.
S24 Cybersecurity and Infrastructure Security Agency. 2025.
S12 ISO/IEC. 2023.
R29 MDPI; industry inspection literature. Publication date unrecorded; inherited source-text check.
S22 National Institute of Standards and Technology. 2023.
S13 ISO/IEC. 2023.
S20 arXiv, author version of an IEEE Access article. 2022-05-14.
S17 Microsoft Learn, Azure Machine Learning. 2026-01-27.
S28 International Society of Automation. 2026.
S40 International Labour Organization and UNESCO. 2020.
S21 arXiv, author manuscript of an IEEE Transactions on Knowledge and Data Engineering article. 2020-04-13.
R27 McKinsey, The State of Fashion 2025. Publication date unrecorded; inherited source-text check.
R12 Business in Cameroon. Publication date unrecorded; inherited source-text check.
Keep the thread · chapter 14 of 14
Move the shuttle along the rack to preview any chapter: its sections, exhibits and reading time. Draft web edition. This research is not investment, legal or tax advice; illustrative models are not measured factory results.
The paper's rack · 14 boltsBolt length: reading time
01Part I: Govern meaning before scale
The binding constraint on industrial AI is often the broken chain between work, record and decision, not the absence of an algorithm.
02Part I: Govern meaning before scale
A workable governance model fits into existing factory roles and makes every definition, exception and release somebody's explicit decision.
03Part I: Govern meaning before scale
Shared identifiers, event semantics and KPI formulas let unlike systems describe the same work without hiding their differences.
04Part I: Govern meaning before scale
A small set of enforced checks, visible exceptions and complete lineage is more useful than an unmeasured claim that the data are clean.
05Part II: Build the minimum durable spine
A durable architecture preserves source evidence, adds governed context once and lets each decision consume a controlled product.
06Part II: Build the minimum durable spine
A value becomes useful evidence only when the factory can establish what it describes, when it happened and which state or transformation it represents.
07Part II: Build the minimum durable spine
The data spine must protect production, people and legitimate access at the same time, with controls proportionate to a small team.
08Part II: Build the minimum durable spine
A factory proves resilience by operating through disconnection, restoring its records and leaving a vendor without losing its history.
09Part III: Control models as production change
The first model gate is operational and economic: nobody should train before the action, baseline, data rights and stop conditions are explicit.
10Part III: Control models as production change
A model becomes a release candidate only when another competent person can reconstruct what data, code, configuration and evaluation produced it.
11Part III: Control models as production change
Release authority follows demonstrated local fitness, safe fallback and accountable acceptance, not a strong aggregate validation score.
12Part III: Control models as production change
A useful monitor detects change, establishes whether it matters and routes a named person through a rehearsed response.
13Part IV: Prove value and scale deliberately
The spine earns further investment when people use it, decisions improve and net operational benefit survives a comparison with the baseline.
14Part IV: Prove value and scale deliberately
GFT should scale the spine only as ownership, meaning, recovery, reproducibility and net value are demonstrated in production.