- UML diagram based on the requirements: diagram.pdf generated from diagram.md
- Custom vocabulary: supply-chain-vocab.ttl
- Shapes: shapes.ttl
- Navigate to
docs. - Install dependencies via
npm i. - Run Walder via
npm run debug. - Navigate to http://localhost:3000/vocab or http://localhost:3000/shapes.
We created the data model as follows:
- We inspected the requirements, previously gathered from the partners, and visualized the possible classes and properties in a diagram that can be found in diagram.pdf.
- We looked at existing vocabularies, especially within the supply chain management, that have the classes and properties that we acquired via the requirements. Grubnic and Fan1 highlighted the gaps in existing vocabularies targeting supply chain management, such as the granularity is only at the strategic level and a restricted view on a supply chain. The Supply Chain Reference Ontology2 (SCRO) takes these gaps into account, while building on top of the core ontology by the Industrial Ontology Foundry (IOF Core). We were able to reuse some of SCRO's and IOF Core's classes, but we also had to look at vocabularies outside the supply chain management domain. Schema.org, and FRAPO are vocabularies that we could reuse and that complement SCRO. After taking these three vocabularies into account, we still did not have all needed classes and properties.
- We created a custom vocabulary, focussed on supply chain management, with the missing classes and properties. See supply-chain-vocab.ttl for details.
- Based on the requirements and the vocabularies we created SHACL shapes that state how the classes and properties should be used together, covering the different use case scenarios. See shapes.ttl for details.
On a high level, we provide a model combining vocabularies and SHACL shapes to description the following modules:
There are three agents defined: a Vendor, a Supplier, and a Customer.
The supply chain consists of Locations, and connections between those locations: LocationToLocation. Each Location serves possible Customers and possible Suppliers, and has an extensible list of options (InventoryHoldingCosts being one of them). Each LocationToLocation has a set of replenishmentMethods.
An Item is within a Vendor’s portfolio. An Item’s Quotation consists of its lead time, distance, and quantity. An Item can have a ML model attached, either a forecast model depending on the Customer, or a prediction model depending on the Supplier.
When a Customer places an Order on an Item, OrderItem metadata is retained, specifically concerning the quantity, and potential order Feedback, i.e., actually delivered quantity, and actual delivery lead time.