Model analisis rangkaian pembelajaran sosial menggunakan teknik pengelompokan ontologi dan ciri-ciri pembelajaran bermakna

Abstract

Clustering on social learning network in e-learning has not been widely explored. Social network analysis requires content analysis involving human intervention which has to be conducted manually. The existing clustering techniques using K-mean and Fuzzy C-mean will determine the centroid for the cluster initialization and these techniques are not suitable for the e-learning data inside the social learning network. This condition happens because the social network analysis method cannot handle large data because these data do not have centroids. This thesis proposes a social learning network analysis model to analyse e-learning data consisting of student activities in an e-learning system. This model integrates the clustering techniques, ontology and meaningful learning characteristics. The clustering process starts by calculating the semantic ontology similarity scores between Moodle e-learning activities and meaningful learning characteristics. Next, it continues by multiplying the hits based on students' actions with the semantic ontology similarity score to gain the cluster score. The cluster scores are categorized according to five meaningful learning characteristics, namely: intentional, active, constructive, cooperative and authentic. This model is further evaluated by comparing the clustering results with the overall marks of the students enrolled in the course. The model was experimented on three samples of e-learning data for three different courses. Prior to performing the t-test, the data were tested and the results showed that the data have fulfilled the normal distribution criteria. Following that, the t-test was applied to all data with H0 representing students' marks on intentional cluster that obtained less mark. The results of the experiment showed that students who belonged to the intentional cluster obtained higher average marks than the other groups. In conclusion, the evaluation results showed that H0 is rejected thus the proposed model is valid with a 95 percent confidence level. This thesis has presented a new model to analyze automatically large amount of e-learning data of students who participated in social learning networks.