Towards a machine learning-based constructive alignment approach for improving outcomes composure of engineering curriculum

• Published in: Education and information technologies Vol. 29; no. 7; pp. 8925 - 8959
• Main Authors: Chor, Wai Tong, Goh, Kam Meng, Lim, Li Li, Lum, Kin Yun, Chiew, Tsung Heng
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.05.2024; Springer; Springer Nature B.V
• Subjects: Affective Objectives; Artificial Intelligence; Behavioral Objectives; Bloom, Benjamin S; Computational linguistics; Computer Appl. in Social and Behavioral Sciences; Computer Science; Computers and Education; Course Content; Curricula; Education; Educational evaluation; Educational Technology; Engineering; Human error; Information Systems Applications (incl.Internet); Language Processing; Learning Activities; Machine learning; Natural language interfaces; Semantics; Support vector machines; Taxonomy; Teaching Methods; User Interfaces and Human Computer Interaction; Malaysia; Engineering Curriculum; Educational Taxonomy; Machine Learning; Course Outcome; Natural Language Processing; Programme Outcome
• ISSN: 1360-2357; 1573-7608
• DOI: 10.1007/s10639-023-12180-y

The programme outcomes are broad statements of knowledge, skills, and competencies that the students should be able to demonstrate upon graduation from a programme, while the Educational Taxonomy classifies learning objectives into different domains. The precise mapping of a course outcomes to the programme outcome and the educational taxonomy (Cognitive, Psychomotor and Affective) level is crucial to ensure Constructive Alignment at the fundamental level of a course and to ensure meaningful outcome measurements. Unfortunately, this effort is often subject to bias and human error while the use of information technologies as a mediator in this area remains unexplored. This research paper proposes an automatic learning-based advisory system for engineering curriculum to ensure constructive alignment with programme outcomes and educational taxonomy. We demonstrated the use of natural language processing and machine learning techniques to mitigate human error and bias that is often present in such classification tasks. Textual/semantic embeddings, including Term Frequency–Inverse Document Frequency (TF-IDF), Universal Sentence Encoder (USE), and Word2Vec (W2V), machine learning models (Random Forest, Support Vector Machine, Logistic Regression, and Light Gradient Boosting Machine), and their corresponding techniques for optimizing the training process are extensively investigated. In terms of accuracy, we obtained an encouraging result of 78.83%, and 78.71% for TF-IDF with Random Forest, and USE with Support Vector Machine classifier, respectively. We transformed our work into a web-based solution named the Course Outcomes Diagnostic Tool, embedded in the faculty education web platform, Edu Centre that is ubiquitously adopted by the members in the Faculty of Engineering and Technology, Tunku Abdul Rahman University of Management and Technology. The proposed solution has demonstrated great potential in reducing subjectivity, ambiguity, and human error, thereby improving the constructive alignment at the root level of course design to ensures teaching–learning activities are aligned with regulatory body expectations.