Semantic model for mining e-learning usage with ontology and meaningful learning characteristics

• Main Author: Octaviani, Dewi
• Format: Thesis
• Language: English
• Published: 2018
• Subjects: LB2300 Higher Education; QA75 Electronic computers. Computer science
• Online Access: http://eprints.utm.my/id/eprint/98239/1/DewiOctavianiPSC2018.pdf; http://eprints.utm.my/id/eprint/98239/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:141931
• Institution: Universiti Teknologi Malaysia

The use of e-learning in higher education institutions is a necessity in the learning process. E-learning accumulates vast amount of usage data which could produce a new knowledge and useful for educators. The demand to gain knowledge from e-learning usage data requires a correct mechanism to extract exact information. Current models for mining e-learning usage have focused on the activities usage but ignored the actions usage. In addition, the models lack the ability to incorporate learning pedagogy, leading to a semantic gap to annotate mining data towards education domain. The other issue raised is the absence of usage recommendation that refers to result of data mining task. This research proposes a semantic model for mining e-learning usage with ontology and meaningful learning characteristics. The model starts by preparing data including activity and action hits. The next step is to calculate meaningful hits which categorized into five namely active, cooperative, constructive, authentic, and intentional. The process continues to apply K-means clustering analysis to group usage data into three clusters. Lastly, the usage data is mapped into ontology and the ontology manager generates the meaningful usage cluster and usage recommendation. The model was experimented with three datasets of distinct courses and evaluated by mapping against the student learning outcomes of the courses. The results showed that there is a positive relationship between meaningful hits and learning outcomes, and there is a positive relationship between meaningful usage cluster and learning outcomes. It can be concluded that the proposed semantic model is valid with 95% of confidence level. This model is capable to mine and gain insight into e-learning usage data and to provide usage recommendation.