OntoLearn: A Formal Ontology for Competency-Based Learning Analytics

Research artifacts for the paper "OntoLearn: A Formal Ontology for Competency-Based Learning Analytics and Semantic Modeling" (SBIE 2026) and the associated PhD thesis (CIn/UFPE).

OntoLearn is an OWL 2 DL ontology (ALCHIF(D) expressivity, as classified by Protégé) for modeling the Knowledge, Skills, and Attitudes (KSA/CHA) triad, proficiency levels, and prerequisite chains in competency-based Learning Analytics.

Repository contents

Folder	Contents
ontology/	ontolearn.owl — final version (v10, OWL/RDF-XML), validated with HermiT 1.4.3 and OOPS!
ontology/versions/	Milestone versions documenting the refinement trajectory (v0, v2, v6)
competency-questions/	The 8 Competency Questions (CQs) used as functional requirements
sparql/	Final SPARQL queries answering each CQ (cq1.rq … cq8.rq)
validation/	Prompt template used with Gemini 3 (PT-BR original + EN translation) and validation notes
slr-protocol/	Systematic Literature Review protocol (databases, search strings, inclusion/exclusion criteria, QA)

How to use

1. Open ontology/ontolearn.owl in Protégé 5.5+.
2. Activate the HermiT reasoner (Reasoner → HermiT → Start reasoner) to materialize inferences (e.g., transitive prerequisite chains via :requer).
3. Run the queries in sparql/ using the SPARQL Query tab (results depend on inferred knowledge — keep the reasoner active).

Base namespace: http://www.ontolearn.com/ontologia#

Version evolution

The ontology was consolidated through iterative OOPS!/SAMOD refinement cycles. Milestone metrics (from Protégé):

Metric	v0 (initial)	v2	v6 (post-OOPS!)	v10 (final)
Classes	82	55	28	29
Object properties	—	26	48	48
Data properties	—	0	27	27
Inverse properties	—	4	24	24
Transitive properties	—	1	2	2
Individuals (test ABox)	0	0	32	40

v0 enumerated entire knowledge domains and ISCED education levels as classes; successive cycles (OOPS! pitfalls P04, P08, P11, P13, P20, P41) replaced this enumerative taxonomy with a leaner, domain-agnostic relational core. Milestone files are in ontology/versions/.

Key modeling notes

• :requer (Competencia → Competencia) is declared owl:TransitiveProperty; :ehRequeridaPor is its declared inverse (also transitive). In the paper, :ehRequeridaPor is presented as isPrerequisiteOf for readability.
• KSA disjointness is enforced pairwise: Knowledge ⊓ Skill ⊑ ⊥, Knowledge ⊓ Attitude ⊑ ⊥, Skill ⊓ Attitude ⊑ ⊥.
• The ABox contains only synthetic instances generated from the schema's constraints. No real student data is included.

Citation

If you use these artifacts, please cite the SBIE 2026 paper:

Nogueira, T. J. D. D.; Lima, R. M. F.; Nogueira, J. C. A. (2026). OntoLearn: A Formal Ontology for Competency-Based Learning Analytics and Semantic Modeling. In: Anais do XXXVII Simpósio Brasileiro de Informática na Educação (SBIE 2026). SBC.

License

• Ontology and datasets: CC BY 4.0
• Documentation: CC BY 4.0

AI Disclosure

Google Gemini 3 (extended-reasoning mode, accessed November 2025) was used as a methodological assistant for translating CQs into SPARQL and for semantic plausibility assessment of axioms, under a human-in-the-loop paradigm. All outputs were executed, verified, and corrected by the authors. See validation/prompt-template.md.