Ontology-Driven ERP Generation: OntoL Turns Business Rules into Executable Systems
This article details how the OntoL ontology platform enables ERP system generation by capturing business entities, relationships, rules, actions, and metric calibers in a formal ontology, allowing LLM-assisted category discovery and automatic translation of business definitions into executable ERP modules without code changes.
LLM-Assisted Business Category Discovery
The process begins with a large language model automatically analyzing an ERP database schema and generating high-level business categories: master data, finance, sales, procurement, inventory, and production. Category colors are assigned automatically, though manual refinement per project is recommended.
Automatic Ontology Generation from Database Scope
After selecting the target database scope, the platform auto-generates the ontology in approximately five minutes. If the database scope is unclear or data governance is lacking, the OntoL-Data suite can integrate unstructured data and heterogeneous system interfaces into a unified high-quality dataset.
Business-Side Complete Generation Path
1. Organize Business Entities (Classes in Ontology)
Business stakeholders inventory all master data and transactional documents in the ERP:
Master data: material, customer, supplier, department, personnel, warehouse, location, operation, work center, account subject, cost center, shift, version (ECN)
Business documents: sales order, delivery order, purchase request, purchase order, receipt order, inspection order, putaway order, issue order, production order, work report, return order, count order, payable order, receivable order, voucher
Ontology role: unify terminology. For example, some departments call it "goods" others "material"; ontology forces a single term material and defines its business attributes, preventing cross-module definition conflicts.
2. Define Business Relationships (Semantic Links, Not Foreign Keys)
Business answers questions that form the process skeleton:
Which customer owns a sales order? Which material does an order line reference?
Which materials (BOM) and operations (routing) does a work order consume?
Which purchase order does a receipt reference? Which business document generated a voucher?
Who owns substitute materials, outsourcing, customer pricing, supplier quotations?
These are business-meaning relationships : A "consumes" B, A "produces" B, A "references" B. The ERP later auto-populates reference lookups and auto-joins based on these semantics.
3. Create Scenario Copies for Role-Based Views
A copy isolates changes from production data. Scenarios define different languages, entity visibility, and field-level permissions per role.
4. Configure Prompt Translator and Data Sources per Scenario
The prompt translator lets different roles see different entity names. Each scenario binds its own data source, enabling conversational querying within permission boundaries.
5. Input Business Rules via Dialogue (Core Control Logic)
Business states operational constraints in natural language; the system generates ERP logic that executes them automatically without developer intervention. Example rules:
Customer credit limit: outstanding + new order amount > limit → order review blocked
Material batch management: outbound must specify batch, FIFO enforced
BOM versioning: ECN effective date locks old BOM; work orders cannot select obsolete version
Inbound requires inspection; failed items cannot enter qualified warehouse
Material issuance cannot exceed work order quota (configurable over-issue tolerance)
Inventory count adjustments require approval before stock update
Core value: business maintains rules; rule changes require no code changes .
6. Define Business Actions (Executable Operations on Documents)
For each document type, business lists allowed operations, preconditions, and resulting data changes:
Sales order: create, approve, lock, change, close, cancel
Work order: release, issue material, report completion, putaway, rework, close
Procurement: request, order, receive, reconcile, invoice, pay
Actions must specify which data fields change: inventory, accounts receivable, accounts payable, work-in-process, available quantity.
7. Define Business Metrics & Calibers (Lock Down Definitions)
Metrics are not arbitrary formulas; the ontology freezes the business caliber so every report shows the same number. Common ERP pain point: sales revenue ≠ finance revenue; ontology locks the caliber.
Metric Catalog by Domain
Sales
Orders: quantity, amount, shipped qty, undelivered qty, overdue orders, on-time delivery rate
Revenue: invoiced amount, uninvoiced revenue, collected amount, collection rate, AR balance, aging
Customer: credit utilization, gross margin, order frequency, return rate, complaint rate
Price: unit price, discount, price change history
Procurement & Supplier
Procurement: request qty, order qty, received qty, in-transit qty, on-time receipt rate, defect rate
Supplier: on-time delivery rate, pass rate, reconciliation cycle, AP balance, purchase price
Cost: price variance, substitute material price difference
Inventory (Most Prone to Caliber Chaos)
Stock: on-hand, available, frozen, in-inspection, in-production, in-transit
Turnover: turnover days, aging, obsolete qty/value, scrap qty
Batch: batch balance, FIFO execution status
Production (Discrete Manufacturing Core)
Work order: released qty, completed qty, scrap qty, rework qty, closure rate
Capacity: reported hours per operation, hour achievement rate, equipment utilization, yield
BOM: loss rate, standard consumption, over-issue qty, substitute usage
MRP: gross requirements, net requirements, planned orders, lead time, shortage alerts
Cost (Finance-Business Integration)
Material, labor, overhead, work order unit cost, completion cost, WIP cost
Variances: material price/quantity, work order variance, scrap loss
Finance
AR, AP, advance receipts, advance payments, GL balances, revenue, cost, expense, profit.
Quality
Incoming IQC defects, in-process IPQC defects, final FQC defects, return defects, rework/repair loss.
Supply Chain Collaboration
Kit completeness rate, shortage count, on-time full-kit delivery rate.
8. Metric Caliber Rules (Five Mandatory Questions per Metric)
Every metric must answer these five questions, stored in the ontology:
Statistical scope : which documents included? (e.g., does revenue include returns? uninvoiced?)
Time dimension : order date / ship date / invoice date?
Filter conditions : exclude voided/cancelled documents?
Calculation formula (business language, not SQL): e.g., on-time delivery rate = on-time delivered orders ÷ due orders
Attribution dimension : split by customer / product line / warehouse / workshop / salesperson
Ontology value: binds this caliber to the metric entity; any report or dashboard pulling the metric automatically follows the same rules — no more one-report-one-algorithm.
Business Boundary (Critical)
Ontology-generated ERP ≠ AI automatically understands business. Prerequisite: business must fully define entities, relationships, process rules, and metric calibers. Ontology is the business "standardized container"; the generator merely translates these definitions into ERP documents, flows, and calculation logic. If business defines incorrectly, ERP logic is wrong; if rules are incomplete, ERP lacks controls.
Business-Side Implementation Sequence (Simplified for Business Owners)
Align master data across sales, procurement, warehouse, production, finance
Map document flow relationships (order → receipt → production → financial voucher)
Extract hard control rules from existing regulations (blocks, alerts, approvals)
Define all daily metrics and unify each metric's statistical caliber
Enter, review, and consistency-check in ontology platform (e.g., detect BOM cycles, credit rule conflicts)
One-click generate ERP business model, pilot; future changes only modify ontology rules, no code changes
Additional Platform Capabilities
OntoL-Manage supports automatic data source configuration, interface configuration, and various full-automation generation features. For organizations without data governance, OntoL-Data provides a middleware layer to consolidate business data, as illustrated:
If users are unsure where to start, OntoL also offers an intermediate warehouse system to help organize business:
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
