Alzet Rama, Nizwardi Jalinus, Muhammad Anwar, Ifdil, Wiki Lofandri
Accessing relevant and context-specific scientific literature in vocational education remains a challenge. Generic academic search engines such as Google Scholar and Semantic Scholar often lack domain-sensitive filters aligned with Technical and Vocational Education and Training (TVET) needs. This study introduces VocScholar, an AI-driven platform tailored to improve the discoverability and relevance of vocational research literature, supporting evidence-based practice and policy in TVET.VocScholar integrates transformer-based semantic search (Sentence-BERT), automated summarization, and a novel Vocational Relevance Score (VRS) to enhance search precision. The platform also includes a dynamic taxonomy mapping engine based on national and ASEAN TVET frameworks, enabling contextual classification across domains such as culinary, automotive, animation, and renewable energy. In comparative testing with 50 vocational educators, VocScholar improved research relevance by 43 % compared to Google Scholar. The system achieved a System Usability Scale (SUS) score of 87,2, indicating high usability. A real-time research dashboard enabled users to explore geospatial and thematic trends in Indonesian and Southeast Asian vocational research. VocScholar narrows the gap between research and vocational practice through intelligent retrieval and domain-aware indexing. It supports curriculum alignment, cross-institutional collaboration, and national innovation agendas. Future work includes multilingual expansion and deeper integration with policy frameworks such as MBKM and UNESCO-UNEVOC’s digital transformation strategy. © 2025; Los autores.
Universitas Negeri Padang, Faculty of Engineering, West Sumatra, Indonesia; Center for Educational Neuroscience, Trauma, and Human Behavior, Universitas Negeri Padang, Indonesia; Universitas Negeri Padang, Faculty of Vocational, West Sumatra, Indonesia
