AI curriculum
Course outline for Knowledge Engineering Foundations for AI Systems
Turn a domain description into information that people and software interpret consistently. Define entities, relationships, categories, and rules. Distinguish what a source says from what follows from accepted facts. Answer new questions and maintain the model when meanings or evidence change. Apply these foundations to reviewing model-extracted claims only after the representation is clear.
Knowledge Engineering Foundations for AI Systems Course Objectives
- Define entities, stable identities, relationships, and a shared glossary.
- Separate categories, instance assertions, ontology commitments, and inference rules.
- Compare document lookup with structured relationship questions.
- Apply bounded explicit rules and show their premises and conclusions.
- Distinguish supported, inferred, disputed, and unknown answers.
- Manage evidence correction, term changes, review ownership, and downstream impact.
Pre-requisites
- Read tables and short documents with potentially conflicting statements.
- Describe a familiar domain without assuming specialist vocabulary.
- Use a spreadsheet or diagram editor; programming is optional.
- No ontology language, graph database, or model subscription is required.
Before this course
Complete the required courses or bring equivalent experience. Recommended courses add useful depth; they are not enrolment gates.
No prerequisite course is required. The entry skills above still apply.
Readiness checks
- Read tables and short documents with potentially conflicting statements.
- Describe a familiar domain without assuming specialist vocabulary.
- Use a spreadsheet or diagram editor; programming is optional.
- No ontology language, graph database, or model subscription is required.
Read tables and short documents with potentially conflicting statements.
Find your place in the learning mapLab Setup
Delivery requirements: the environment, access, and exercise materials must be prepared and verified before delivery. This outline does not certify that a lab package is ready.
Spreadsheet or diagram editor, synthetic library records, lending policy, glossary starter, rule table, contradictory evidence, and separate explanatory feedback. Materials include document-lookup and structured-query examples. Delivery requires an independently checked rule procedure and answer key. No programming runtime, ontology reasoner, graph/vector service, or model subscription is required.
Detailed Course Outline
Module 1: Give things and relationships a shared meaning (2 hours)
- Representation choices: Representation choices: documents, tables, diagrams, and structured assertions.
- Entities and identity: Entities, identifiers, names, attributes, and relationship direction.
- Sourced assertions: Facts as assertions with sources rather than unquestionable truth.
- Shared glossary: Glossaries, ambiguous terms, domain boundaries, and competency questions.
- Categories and taxonomies: Categories, taxonomy hierarchies, membership, and part-versus-kind relationships.
- Schema and instances: Schema definitions versus individual instances and their evidence.
- Guided-time allocation: 0.75 hours for explanation and a worked example, 0.75 hours for learner practice, 0.25 hours for feedback, and 0.25 hours for questions and contingency.
Practical outcome: Model a small lending library and answer which meanings a plain document leaves ambiguous.
Module 2: Explain what follows from the model (2 hours)
- Ontology commitments: Ontologies as explicit domain commitments beyond category lists.
- Semantic models: Semantic models connecting shared terms, identities, and relationship meanings.
- Rule inference: Rule premises, conclusions, conjunction, and a bounded inference procedure.
- Asserted and inferred answers: Asserted versus inferred answers and traceable reasoning steps.
- Unknown and disputed claims: Unknown, disputed, and explicitly negative claims with defined semantics.
- Retrieval and relationship queries: Document retrieval versus relationship queries for the same question.
- Guided-time allocation: 0.75 hours for explanation and a worked example, 0.75 hours for learner practice, 0.25 hours for feedback, and 0.25 hours for questions and contingency.
Practical outcome: Derive a lending-eligibility result and identify one answer the available evidence cannot establish.
Module 3: Keep knowledge useful as evidence changes (2 hours)
- Knowledge lifecycle: Acquisition, review, publication, use, correction, and retirement.
- Evidence and validity: Source authority, validity periods, version history, and conflicting evidence.
- Definition changes: Changes to definitions versus changes to instance assertions.
- Correction impact: A correction's impact on inferred answers and downstream users.
- Extraction review: Model-extracted candidates, evidence review, and unsupported assertions.
- Stewardship and simpler alternatives: Maintenance ownership and when a maintained document is sufficient.
- Guided-time allocation: 0.75 hours for explanation and a worked example, 0.75 hours for learner practice, 0.25 hours for feedback, and 0.25 hours for questions and contingency.
Practical outcome: Apply a correction without erasing history and review an AI-extracted candidate against its source.
Practical exercise
- Model library members, items, copies, loans, and lending categories. Start with four questions about availability and eligibility. Define identifiers and a glossary. Separate an item from its physical copy and a category from its members. Create a taxonomy and a small ontology diagram with defined relation meanings. Use the supplied lending policy before applying inference rules.
- Apply explicit supplied rules to learner-authored assertions. Use a spreadsheet rule table or a fully specified manual procedure. Show each premise and derive one answer through two rule steps. Predict the answer to a new case before opening explanatory feedback. A mechanical rule procedure is sufficient here; formal reasoner operation belongs in Semantic Web and Ontology Engineering. Distinguish unknown from false under the stated rule policy. Record contradictory assertions without using both as unqualified accepted premises.
- Compare a source passage with a structured relationship answer. Correct one loan record and one term definition. Identify affected facts, inferred answers, and consumers, preserving prior versions. Finally review supplied model-extracted claims: accept a supported relation, reject a fabricated one, and leave an ambiguous reference unresolved. No live-model or automated extraction capability is claimed. Provenance records origin, not guaranteed truth.
Delivery and timing
- Guided hours include teaching, demonstrations, learner practice, feedback, questions, and contingency as allocated above. Breaks, bulk downloads, and prerequisite setup are additional. Allocations need a timed learner pilot before delivery; they are not measured completion times.
What you will produce
Build and maintain a shared knowledge model with explicit meanings, evidence, rules, and correction ownership.
How the work is checked
- Identities, meanings, category relationships, and instance assertions remain distinct.
- The model answers its competency questions or explains missing information.
- Inferences show accepted premises and correct rule application.
- Unknown and disputed claims do not silently become negative or accepted facts.
- Corrections identify affected conclusions and preserve ownership and history.
- Extraction review checks source evidence rather than confidence or fluent wording.
6 guided hours. General knowledge engineering comes first; supplied model-extraction review is supporting transfer. A spreadsheet or fully specified manual inference procedure is sufficient. Formal reasoner operation, RAG construction, graph-database implementation, and live-model extraction are not claimed. Prefer stable documents and a glossary when they answer the questions.
When another approach is better
- A maintained document may suffice when relationships are rarely queried or reused.
- Do not freeze an ontology before domain terms and source ownership are understood.
How we train
Contact us for full course details, including duration, delivery options and lab requirements.
Ask about exercises, instructor feedback and the prior knowledge you need. Tool-specific courses marked provisional may become modules in a broader course.
Discuss your learning goalsRelated Courses
These courses list this course as preparation.
Courses that require this course
- Semantic Web and Ontology Engineering
- Knowledge Graph Engineering
- Context Engineering and Context Graphs
- Neuro-Symbolic AI