Online Reputation of Malaysian Technical and Vocational Education: A Context-Aware Analysis

Technical and vocational education and training (TVET) is increasingly positioned as a strategic pathway for workforce development, yet its public reputation is formed through fragmented news and social-media discourse. This study develops and applies a two-layer context-aware procedure to identify corporate-reputation dimensions, sentiment and potential topics in Malaysian TVET discussion. Layer 1 generated R-PMI-BERT features from RepLab 2014 by combining normalised pointwise-mutual-information relevance with BERT contextual embeddings; the benchmark achieved 0.87 accuracy and a macro F1-score of 0.80. The feature inventory was then applied to online discourse of TVET in Malaysia published between 2 January 2024 and 10 June 2026. Performance and Citizenship were the most frequent dimensions, each accounting for 22.9% of the 83 parent records, followed by Workplace (21.7%), Innovation (13.3%), Leadership (12.0%), and Governance (7.2%). Layer 2 sentiment aggregation classified 58 records as positive (69.9%), 16 as neutral (19.3%) and nine as negative (10.8%). Performance had the highest concentration of positive records, while the negative records were distributed across Citizenship, Innovation, Leadership, Performance and Workplace. The emerging narrative emphasised employability, access, industry alignment, reform, funding, technological transition, certification and career pathways. The results indicate a broadly favourable but dimensionally uneven reputation centred on societal outcomes and labour-market relevance. Nevertheless, the use of cross-domain feature transfer, weak dimension labels and lexicon-based sentiment require subsequent human validation.