Noetva

Sample Dataset

Architecture

This page explains, in business terms, how the sample datasets behind the Supplier Risk demo become an explainable recommendation. It describes the working demo — not a live enterprise system.

The five-stage journey

How a sample dataset becomes a recommendation

  1. Import Data

    The seven controlled sample CSV files are parsed into the demo's known dataset structure.

    These are fixed demo fixtures, not live connectors or enterprise data sources:

    • suppliers.csv
    • materials.csv
    • products.csv
    • facilities.csv
    • supply_agreements.csv
    • risk_events.csv
    • revenue_exposure.csv

  2. Map Schema

    Technical fields are mapped to business concepts using predefined sample mappings for this demo.

    Each sample mapping includes a predefined sample mapping confidence value shown for illustration — it is fixed demo metadata, not a dynamically calculated or inferred confidence score.

  3. Explore Ontology

    Records are connected into the sample ontology shown below.

    Constructing these relationships is what enables impact analysis — calculating which materials, products, facilities, and revenue are connected to a disruption. Impact analysis is a capability within this stage, not a separate stage.

  4. Decision Flow

    Explicit, deterministic rules are evaluated against the resolved sample facts.

    No AI, LLM, or autonomous agent makes this evaluation — the conditions and their outcomes are plain, inspectable logic.

  5. Recommendation

    The demo proposes a recommendation for human review.

    Approve or Reject records the reviewer's decision in the current demo session. No operational action is executed and no enterprise system is updated. Recommendation factors and rule conditions are traceable to sample source rows — a fixture filename and row index — which are demo fixture references, not persisted enterprise lineage or immutable evidence records.

Sample Supplier Risk Relationships

These are the relationships supported by the current sample ontology used in the Supplier Risk demo. This is not a universal ontology, a comprehensive enterprise model, or an automatically discovered semantic graph.

  • Supplier —supplies→ Material
  • Material —used in→ Product
  • Product —assembled at→ Facility
  • Facility —generates revenue→ Revenue Exposure
  • Supplier —located in→ Region
  • Region —affected by→ Risk Event
  • Material —covered by→ Supply Agreement
  • Candidate Supplier —candidate alternate for→ Material

Scope

What this demo is and isn't

"Architectural direction" describes where this approach could go — it is not already available in this demo.

Implemented in this demo

  • Parsing the seven controlled sample CSV files into the demo's known dataset structure
  • Duplicate identifier and broken reference detection within the sample dataset
  • Constructing the sample ontology from resolved sample facts
  • Calculating impact (revenue exposure, affected counts) from the sample ontology
  • Evaluating explicit, deterministic decision rules
  • Recording an Approve or Reject decision in the current demo session

Architectural direction

  • Applying this same layered approach to an organization's own data
  • Extending the ontology and decision rules beyond a single sample scenario

Not implemented

  • Live enterprise connectors
  • Persistent decisions
  • Operational execution
  • Production evidence storage
  • Multi-scenario ontology reasoning
  • AI, LLM, or agent decision-making
  • Dynamically calculated confidence
  • Enterprise authentication or authorization
How this demo works

This page is understandable without this section. It's here for anyone curious about how each stage above is actually built.

  • Parses the seven sample CSV files into the demo's known dataset structure. fixture-loader.ts
  • Checks the parsed dataset for duplicate identifiers and broken references. dataset-validation.ts
  • Resolves the sample scenario's connected facts from the dataset, or reports exactly which relationship is missing. scenario-facts.ts
  • Builds the sample ontology graph from the resolved facts. ontology-model.ts
  • Evaluates the deterministic decision rules against the resolved facts. decision-rules.ts