EXPLORING LATENT TOPICS IN TVET USING LDA TOPIC MODELING
Nur Hafazah Sharin1, Mira Kartiwi1
1Kulliyyah of Information and Communication Technology, International Islamic University Malaysia, Malaysia
Social media contains vast amounts of textual data that can assist organizations in understanding their stakeholders better. To study public perceptions of Technical and Vocational Education and Training (TVET) in Malaysia, collecting data from social media is necessary. 1,304 Facebook posts from ministry, news and media pages, and public groups were analyzed. Latent Dirichlet Allocation (LDA) topic modeling was utilized to uncover hidden themes by identifying the number of topics and associated keywords. With the highest coherence value of 0.4717, the analysis extracted eight (8) relevant themes regarding TVET. Ten (10) keywords from each topic help classify the TVET topic. The top three (3) topics that have been classified are skills/ competency, certification, and salary/ wage. By gaining the extracted topics, would assist in decision-making and improve the TVET ecosystem.
TVET, Topic Modeling LDA, Coherence