Interconnectivity and Relationship Modeling

SyntheticData provides comprehensive relationship modeling capabilities for generating realistic enterprise networks with multi-tier vendor relationships, customer segmentation, relationship strength calculations, and cross-process linkages.

Overview

Real enterprise data exhibits complex interconnections between entities:

• Vendors form multi-tier supply chains (supplier-of-supplier)
• Customers segment by value (Enterprise vs. SMB) with different behaviors
• Relationships vary in strength based on transaction history
• Business processes connect (P2P and O2C link through inventory)

SyntheticData models all of these patterns to produce realistic, interconnected data.

Multi-Tier Vendor Networks

Supply Chain Tiers

Vendors are organized into a supply chain hierarchy:

Vendor Clusters

Vendors are classified into behavioral clusters:

Vendor Lifecycle Stages

Onboarding → RampUp → SteadyState → Decline → Terminated

Each stage has associated behaviors:

• Onboarding: Initial qualification, small orders
• RampUp: Increasing order volumes
• SteadyState: Stable, predictable patterns
• Decline: Reduced orders, performance issues
• Terminated: Relationship ended

Vendor Quality Scores

Vendor Concentration Analysis

SyntheticData tracks vendor concentration risks:

dependencies:
  max_single_vendor_concentration: 0.15  # No vendor > 15% of spend
  top_5_concentration: 0.45              # Top 5 vendors < 45% of spend
  single_source_percent: 0.05            # 5% of materials single-sourced

Customer Value Segmentation

Value Segments

Customers follow a Pareto-like distribution:

Customer Lifecycle

Prospect → New → Growth → Mature → AtRisk → Churned
                                         ↓
                                      WonBack

Each stage has associated behaviors:

• Prospect: Potential customer, conversion probability
• New: First purchase within 90 days
• Growth: Increasing order frequency/value
• Mature: Stable, loyal customer
• AtRisk: Declining activity, churn signals
• Churned: No activity for extended period
• WonBack: Previously churned, now returned

Customer Engagement Metrics

Customer Networks

• Referral Networks: Customers refer other customers (configurable rate)
• Corporate Hierarchies: Parent/child company relationships
• Industry Clusters: Customers grouped by industry vertical

Relationship Strength Modeling

Composite Strength Calculation

Relationship strength is computed from multiple factors:

Strength Categories

Recency Decay

The recency component uses exponential decay:

recency_score = exp(-days_since_last / half_life)

Default half-life is 90 days.

Cross-Process Linkages

Inventory Links (P2P ↔ O2C)

Inventory naturally connects Procure-to-Pay and Order-to-Cash:

P2P: Purchase Order → Goods Receipt → Vendor Invoice → Payment
                           ↓
                      [Inventory]
                           ↓
O2C: Sales Order → Delivery → Customer Invoice → Receipt

When enabled, SyntheticData generates explicit CrossProcessLink records connecting:

• GoodsReceipt (P2P) to Delivery (O2C) via inventory item

Payment-Bank Reconciliation

Links payment transactions to bank statement entries for reconciliation.

Intercompany Bilateral Links

Ensures intercompany transactions are properly linked between sending and receiving entities.

Entity Graph

Graph Structure

The EntityGraph provides a unified view of all entity relationships:

Entity Types (16 types)

Company, Vendor, Customer, Employee, Department, CostCenter,
Project, Contract, Asset, BankAccount, Material, Product,
Location, Currency, Account, Entity

Relationship Types (26 types)

// Transactional
BuysFrom, SellsTo, PaysTo, ReceivesFrom, SuppliesTo, OrdersFrom

// Organizational
ReportsTo, Manages, BelongsTo, OwnedBy, PartOf, Contains

// Network
ReferredBy, PartnersWith, AffiliateOf, SubsidiaryOf

// Process
ApprovesFor, AuthorizesFor, ProcessesFor

// Financial
BillsTo, ShipsTo, CollectsFrom, RemitsTo

// Document
ReferencedBy, SupersededBy, AmendedBy, LinkedTo

Configuration

Complete Example

vendor_network:
  enabled: true
  depth: 3
  tiers:
    tier1:
      count_min: 50
      count_max: 100
    tier2:
      count_per_parent_min: 4
      count_per_parent_max: 10
    tier3:
      count_per_parent_min: 2
      count_per_parent_max: 5
  clusters:
    reliable_strategic: 0.20
    standard_operational: 0.50
    transactional: 0.25
    problematic: 0.05
  dependencies:
    max_single_vendor_concentration: 0.15
    top_5_concentration: 0.45
    single_source_percent: 0.05

customer_segmentation:
  enabled: true
  value_segments:
    enterprise:
      revenue_share: 0.40
      customer_share: 0.05
      avg_order_min: 50000.0
    mid_market:
      revenue_share: 0.35
      customer_share: 0.20
      avg_order_min: 5000.0
      avg_order_max: 50000.0
    smb:
      revenue_share: 0.20
      customer_share: 0.50
      avg_order_min: 500.0
      avg_order_max: 5000.0
    consumer:
      revenue_share: 0.05
      customer_share: 0.25
      avg_order_min: 50.0
      avg_order_max: 500.0
  lifecycle:
    prospect_rate: 0.10
    new_rate: 0.15
    growth_rate: 0.20
    mature_rate: 0.35
    at_risk_rate: 0.10
    churned_rate: 0.08
    won_back_rate: 0.02
  networks:
    referrals:
      enabled: true
      referral_rate: 0.15
    corporate_hierarchies:
      enabled: true
      hierarchy_probability: 0.30

relationship_strength:
  enabled: true
  calculation:
    transaction_volume_weight: 0.30
    transaction_count_weight: 0.25
    relationship_duration_weight: 0.20
    recency_weight: 0.15
    mutual_connections_weight: 0.10
    recency_half_life_days: 90
  thresholds:
    strong: 0.7
    moderate: 0.4
    weak: 0.1

cross_process_links:
  enabled: true
  inventory_p2p_o2c: true
  payment_bank_reconciliation: true
  intercompany_bilateral: true

Network Evaluation

SyntheticData includes network metrics evaluation:

These metrics validate that generated networks exhibit realistic properties.

API Usage

Rust API

#![allow(unused)]
fn main() {
use datasynth_core::models::{
    VendorNetwork, VendorCluster, SupplyChainTier,
    SegmentedCustomerPool, CustomerValueSegment,
    EntityGraph, RelationshipStrengthCalculator,
};
use datasynth_generators::relationships::EntityGraphGenerator;

// Generate vendor network
let vendor_generator = VendorGenerator::new(config);
let vendor_network = vendor_generator.generate_vendor_network("C001");

// Generate segmented customers
let customer_generator = CustomerGenerator::new(config);
let customer_pool = customer_generator.generate_segmented_pool("C001");

// Build entity graph with cross-process links
let graph_generator = EntityGraphGenerator::with_defaults();
let entity_graph = graph_generator.generate_entity_graph(
    &vendor_network,
    &customer_pool,
    &transactions,
    &document_flows,
);
}

Python API

from datasynth_py import DataSynth
from datasynth_py.config import VendorNetworkConfig, CustomerSegmentationConfig

config = Config(
    vendor_network=VendorNetworkConfig(
        enabled=True,
        depth=3,
        clusters={"reliable_strategic": 0.20, "standard_operational": 0.50},
    ),
    customer_segmentation=CustomerSegmentationConfig(
        enabled=True,
        value_segments={
            "enterprise": {"revenue_share": 0.40, "customer_share": 0.05},
            "mid_market": {"revenue_share": 0.35, "customer_share": 0.20},
        },
    ),
)

result = DataSynth().generate(config=config, output={"format": "csv"})

See Also

• Graph Export - Exporting entity graphs to PyTorch Geometric, Neo4j, DGL
• Intercompany Processing - Multi-entity transaction matching
• Master Data Configuration - Vendor and customer settings