Model enterprise entities and relationships as triples and choose an appropriate graph representation.
Digital Transformation & Artificial Intelligence
Knowledge Graphs and Ontologies for Enterprise AI Applications
Teaches data and AI teams to design ontologies, build enterprise knowledge graphs and use them to ground AI applications in verified, traceable facts.
🗓️Duration5 training days
📚Content4 modules · 8 sessions
🎓On completionAccredited attendance certificate
About the programme
Course Overview
Generative AI applications fail most visibly when they state a fabricated fact with total confidence, and in enterprise AI applications the most reliable fix is giving the model a verified structure to draw from instead of open text alone. This course builds that structure from the ground up: modelling entities and relationships as triples, designing an ontology whose classes and properties reflect how the business actually categorises what it knows, and extracting facts from both structured systems and unstructured documents without breaking the ontology's rules. Entity resolution work covers matching and merging duplicate records across source systems without silently losing relationships in the process. Query and reasoning sessions cover graph query languages and the inference rules that let a graph answer questions it was never explicitly loaded to answer. The final sessions connect the graph to generative AI directly, retrieving verified facts to ground a model's response and tracing every generated answer back to the specific facts that supported it, then set out how an ontology is governed and versioned so it keeps working as the business changes.
Expected Learning Outcomes
Design an ontology with classes, properties and controlled vocabularies reviewed by domain experts.
Extract and validate entities and relationships from structured and unstructured sources against the ontology.
Resolve and merge duplicate entities across systems without losing their accumulated relationships.
Query an enterprise knowledge graph using standard graph query languages with reliable performance.
Apply ontology-based inference to derive new facts and verify them against source data.
Ground generative AI responses in graph-verified facts with a traceable citation for each claim.
Who Should Attend
Data architects designing a knowledge graph to underpin enterprise search and AI applications.
Ontologists and information architects responsible for enterprise taxonomies and vocabularies.
AI engineers building retrieval-augmented generation systems that must cite verified facts.
Master data management specialists resolving duplicate entities across source systems.
Data scientists who need graph-structured features or context for machine learning models.
Enterprise architects evaluating knowledge graphs against relational alternatives for a given use case.
Course Modules
Select any module to see its sessions and points.
01Foundations of Knowledge Graphs and Ontology Design
2 sessions · 8 points
Session 1Modelling Entities, Relationships and Triples
- Model core business entities and their relationships as a set of subject-predicate-object triples.
- Choose between a property graph and a triple-based model based on query patterns and tooling available.
- Identify which existing relational tables describe entities and relationships better suited to a graph representation.
- Sketch a minimum viable graph schema around one business domain before attempting an enterprise-wide model.
Session 2Designing an Ontology With Classes and Controlled Vocabularies
- Define ontology classes and subclasses that reflect how the business actually categorises its entities.
- Specify properties and their allowed value types so the ontology constrains what a valid fact looks like.
- Reuse an established controlled vocabulary where one exists instead of inventing new terms for the same concept.
- Review the draft ontology with domain experts before it is used to structure production data.
02Building and Populating the Graph
2 sessions · 8 points
Session 1Extracting Entities and Relationships From Enterprise Sources
- Extract entities and relationships from structured source systems using their existing keys and foreign keys.
- Apply named entity recognition and relation extraction to pull facts from unstructured text sources.
- Validate automatically extracted relationships against the ontology before loading them into the graph.
- Track the source system and extraction method for every triple so provenance survives into the graph.
Session 2Resolving and Linking Duplicate Entities Across Systems
- Match candidate entities across source systems using shared identifiers, similarity scoring or both.
- Resolve conflicting attribute values for the same real-world entity according to a documented precedence rule.
- Merge duplicate entity nodes without losing the relationships each duplicate had accumulated separately.
- Flag low-confidence entity matches for human review rather than merging them automatically.
03Querying and Reasoning Over Enterprise Knowledge Graphs
2 sessions · 8 points
Session 1Querying Graphs With Standard Graph Query Languages
- Write graph queries that traverse multiple relationship hops to answer questions a relational join struggles with.
- Optimise query patterns and indexing so traversal performance holds up as the graph grows.
- Expose a controlled query interface to business analysts who should not write raw graph query syntax themselves.
- Test queries against known correct answers before trusting graph output in a business report.
Session 2Applying Inference to Derive New Facts From the Ontology
- Apply ontology-based inference to derive relationships that were never explicitly loaded into the graph.
- Use class hierarchies so a query for a general category automatically includes its defined subclasses.
- Check inferred facts against source data periodically to confirm reasoning rules still produce valid conclusions.
- Document each inference rule so analysts can distinguish an asserted fact from a derived one.
04Applying Knowledge Graphs to Enterprise AI
2 sessions · 8 points
Session 1Grounding Generative AI Outputs in a Knowledge Graph
- Retrieve relevant graph facts to ground a generative AI response before it is presented to a user.
- Constrain a language model's answer to entities and relationships that exist in the verified graph.
- Trace a generated answer back to the specific graph facts that supported it for audit purposes.
- Reduce ungrounded, fabricated responses by requiring a graph-verified citation for factual claims.
Session 2Governing and Evolving the Ontology Over Time
- Assign ontology stewardship to a named owner who approves new classes and properties before release.
- Version the ontology so applications built against an earlier version keep working during evolution.
- Review the graph for orphaned nodes and stale relationships as part of a recurring maintenance cycle.
- Measure graph coverage and query success rate to prioritise where ontology investment goes next.
What the participant receives
4 course modules
A structured syllabus
8 training sessions
across 5 days
32 detailed points
Applied, detailed content
Accredited attendance certificate
On completing the programme
Complete your registration
We will contact you within one business day to confirm.
Ready to start?
Reserve your seat and start building the skill.