Uncovering Latent Influences on Student Enrollment in the Human Services Program Using a Hybrid NLP–Fuzzy DEMATEL Approach

Understanding the motivations behind senior high school students’ decisions to pursue a Bachelor of Science in Human Services is essential for improving recruitment strategies and educational policy. This study presents a hybrid methodology integrating Natural Language Processing (NLP), lexicon-based semantic mapping, and an enhanced Fuzzy DEMATEL framework with automated scoring. From 1,054 open-ended survey responses, a bilingual lexicon (English–Cebuano–Tagalog) and Word2Vec embeddings trained on a multilingual, code-switched student corpus were used to compute semantic similarity scores with seed concepts, enabling direct-relation matrix generation without subjective expert input. This automation improves scoring consistency, reduces bias, and strengthens causal interpretation. Benchmarking against human annotations yielded 78% agreement (Cohen’s κ = 0.76), and comparisons with FastText and multilingual BERT confirmed Word2Vec’s effectiveness. Results identified Parental Influence and Scholarship Importance as dominant decision drivers, with Career Opportunities and Personal Interest acting as both influencing and influenced factors, reflecting the interplay between intrinsic motivation and external support. Validation via analogy tests achieved 74% accuracy, semantic coherence ranged from 0.71–0.83, and cross-domain tests reached 65–72% accuracy, indicating reasonable generalizability. By contextualizing decision factors in a culturally and linguistically relevant manner, the study offers actionable insights for targeted scholarships, recruitment, and policy-making grounded in data-driven understanding of student needs.