Skip to content

Concept Taxonomy

12 categories covering all 496 concepts. No category exceeds 14% of total concepts.

TaxonomyID Category Name Concepts % Description
GRAPH Graph Fundamentals 45 9.1% Core graph theory: nodes, edges, LPG model, Cypher, graph algorithms, schema, serialization
EKG Enterprise Knowledge Graphs 30 6.0% Building and scaling LPG-based knowledge graphs across enterprise systems
DATA Data and Semantic Layers 47 9.5% Data lakes, semantic layers, metric stores, metadata management, and data governance
META Metadata Standards 28 5.6% ISO 11179, metadata registries, data element definitions, NIEM, UMLS, thesauri
PROV Process Mining and Provenance 25 5.0% Process mining, event logs, data lineage, data provenance, change data capture
CNTX Context Problem and Context Graphs 40 8.1% The context problem, RAG limitations, context graph definition, schema, API, lifecycle
DTRC Decision Traces and Graph Modeling 66 13.3% Decision trace anatomy, LPG implementation, incumbent challenges, graph data modeling
LLMI LLM Foundations and Integration 45 9.1% Transformer architecture, prompting, fine-tuning, context retrieval, grounding strategies
BUILD Building and Deploying Systems 50 10.1% Storage layers, ingestion pipelines, SDK, monitoring, AI agent architecture
USE Enterprise Use Cases 25 5.0% Finance, sales, engineering, legal, HR, healthcare, and cross-domain applications
COMP Compliance, Market, and Adoption 55 11.1% Audit trails, explainability, regulatory compliance, startup strategies, organizational adoption
INFRA Data Infrastructure and Security 40 8.1% Data mesh, streaming platforms, vector databases, graph security, privacy, federated learning

Category Definitions

GRAPH — Graph Fundamentals

Concepts: 1–25 (Topic 1) and 417–436 (cross-cutting graph theory)

The foundational building blocks of the Labeled Property Graph model and graph theory: nodes, edges, labels, properties, schemas, Cypher/openCypher/GQL query languages, graph algorithms (shortest path, PageRank, community detection), graph schema, serialization formats (GraphML, GraphSON), and the comparison of LPG with relational and RDF models including the scalability limitations of RDF.

EKG — Enterprise Knowledge Graphs

Concepts: 26–55 (Topic 2)

How organizations build and operate LPG-based knowledge graphs at enterprise scale: entity resolution, master data management, canonical entity models, hub-and-spoke and federated architectures, cross-system entity linking, ETL and ingestion patterns, graph sharding and replication, knowledge graph quality, ontologies, SKOS, and graph data catalogs.

DATA — Data and Semantic Layers

Concepts: 56–102 (Topics 3 and 4)

The infrastructure that gives enterprise data meaning: data lakes and lakehouses, semantic layers, metric stores, business glossaries, query federation, data virtualization, metadata management (technical, business, operational), data quality dimensions, data governance frameworks, stewardship, ownership, classification, access control, automated metadata discovery, and metadata catalog platforms.

META — Metadata Standards

Concepts: 103–130 (Topic 5)

Formal standards and vocabularies for enterprise metadata: the ISO/IEC 11179 metadata registry standard and its six core components (data element, data element concept, conceptual domain, value domain, object class, property), registration authorities, naming conventions, reference data management, code lists, metadata thesauri, UMLS, and NIEM.

PROV — Process Mining and Provenance

Concepts: 131–155 (Topic 6)

Reconstructing what actually happened from event logs and streams: process mining (process discovery, conformance checking, process enhancement), the IEEE XES standard, structured logging, upstream and downstream lineage, column-level lineage, data provenance, custody chains, transformation history, the OpenLineage standard, event sourcing, CQRS, append-only logs, change data capture, and temporal versioning.

CNTX — Context Problem and Context Graphs

Concepts: 156–195 (Topics 7 and 8)

The gap that context graphs fill: why LLMs fail at enterprise tasks even with existing infrastructure; RAG and its limits; tacit and implicit knowledge; context freshness, relevance, and poisoning; the definition of a context graph; how it extends rather than replaces the enterprise knowledge graph; context graph schema, API, read/write paths, lifecycle, and access control.

DTRC — Decision Traces and Graph Modeling

Concepts: 196–261 (Topics 9, 10, and 11)

The core of context graph content: the anatomy of a decision trace; exception logic, historical precedent, approval chains, and out-of-band approvals; LPG implementation patterns (decision trace node schema, edge types, trace-to-entity relationships, actor nodes, policy version edges, precedent chains); why incumbent systems will struggle to build this; and the full graph data modeling toolkit for context (bitemporal modeling, temporal edges, valid/transaction time, constraint enforcement, schema evolution).

LLMI — LLM Foundations and Integration

Concepts: 262–286 (Topic 12) and 457–476 (cross-cutting LLM foundations)

How large language models work and how they connect to context graphs: transformer architecture, tokens, prompting, fine-tuning, RLHF, in-context learning, zero- and few-shot prompting; context retrieval, relevance ranking, vector embeddings, hybrid retrieval, context window budgeting, compression and pruning, grounding strategies, hallucination mitigation, function calling, context graph tool definitions, and prompt injection risks.

BUILD — Building and Deploying Systems

Concepts: 287–336 (Topics 13 and 14)

End-to-end system construction: storage layer selection (property graph database, vector index, hybrid), ingestion pipelines (real-time, batch, backfill, event-driven), SDK/REST/GraphQL APIs, caching, replication, monitoring, alerting, SLAs, deployment patterns, testing strategies, observability, cost models; and the AI agent architecture layer: agent loop design, planning, memory, tool use, orchestration, multi-agent systems, graduated autonomy, human-in-the-loop, write-back, ReAct/Plan-and-Execute/ Reflection patterns, and agent evaluation.

USE — Enterprise Use Cases

Concepts: 337–361 (Topic 15)

Concrete applications of context graphs across the enterprise: finance (revenue reporting exceptions, ARR definition conflicts), sales (engagement history, playbook precedents), engineering (incident traces, production decision records), legal/compliance (regulatory audit automation), customer success (account history, escalation logic), HR, procurement, supply chain, marketing, risk, IT operations, healthcare, financial services, retail, manufacturing, insurance, government, and cross-department patterns.

COMP — Compliance, Market, and Adoption

Concepts: 362–416 (Topics 16, 17, and 18)

The organizational and regulatory context for context graphs: GDPR right to explanation, audit trail design, explainability by design vs. post-hoc, algorithmic accountability, bias/fairness audits, data retention and purge policies, compliance reporting, the EU AI Act; startup strategy (full replacement, module replacement, new system); beachhead workflow selection, glue functions, competitive moats; and the organizational adoption lifecycle (first workflow selection, change management, stakeholder alignment, ROI measurement, graduated autonomy rollout, center of excellence).

INFRA — Data Infrastructure and Security

Concepts: 437–456 (cross-cutting data engineering) and 477–496 (security and vector search)

Supporting infrastructure: data mesh, data products and contracts, data SLAs, data observability, SQL transformation tools, workflow orchestration, event streaming platforms, graph batch and streaming processing, multi-model and time-series integration, feature engineering from graphs; graph security models, row-level security, attribute-based access control, zero-trust architecture, data masking, anonymization, differential privacy, federated learning, model audit trails, AI red teaming; vector databases (HNSW, product quantization, ANN), embedding models, sentence transformers, dense and sparse retrieval (BM25), and context graph ROI modeling.