Work

Ontology and Natural Language Processing for Hierarchical Alignment Traceability in Outcome-Based Education

Year: 2026
Type: conference-paper
Abstract: Outcome-Based Education (OBE) demands coherence and consistency within five levels of the curriculum hierarchy: Graduate Profile, Program Learning Outcomes (PLO), Course Learning Outcomes (CLO), Sub-CLO, and Assessment Items. Current automated solutions are working only at three of the five levels, and they do not have an ontological representation and a built-in Natural Language Processing (NLP) pipeline to score the semantics of the sentences, and they do not solve the full five-level traceability problem. This work proposes a Multi-layer Ontological framework for Hierarchical Alignment Traceability (MOHAT) in OBE, which combines a 5-layer OWL 2 DL ontology and an embedded semantic similarity NLP pipeline to achieve automated alignment traceability, spanning all five OBE levels. The six major classes, four object properties, and three SWRL inference rules (transitive traceability and misalignment flagging) constitute the formal representation of curriculum relationships. The NLP pipeline has been divided into four alignment levels based on the cosine similarity scores obtained between adjacent level statement pairs. The framework was tested with an OBE curriculum dataset that contained 56 pairs of experts annotated by agreement $(\boldsymbol{\kappa}=\mathbf{0. 8 1})$; MOHAT-OBE achieved an $\mathbf{F 1}$-Score $\mathbf{= 0. 8 6 0}$ and $\mathbf{A U C}-\mathbf{R O C}=\mathbf{0. 8 8 9}$, followed by TF-IDF $(\boldsymbol{\Delta} \mathbf{F} \mathbf{1}=+\mathbf{0. 1 5 5})$, Word2Vec $(\boldsymbol{\Delta} \mathbf{F} \mathbf{1}=+\mathbf{0. 1 2 5})$, BERT-base $(\boldsymbol{\Delta} \mathbf{F} \mathbf{1}=+\mathbf{0. 0 7 3})$, and RoBERTa $(\boldsymbol{\Delta} \mathbf{F} \mathbf{1}=+\mathbf{0. 0 5 3})$. This framework gives complete 5-level OBE traceability and integrated ontological reasoning and automated NLP scoring and has been proven to be applicable to the Indonesian standard curricula.
Language: English
Open Access status: hybrid
OpenAlex ID: W7225395664
FWCI:
Cites: 20
Cited by:
Related to: 10