Pengembangan Dan Uji Efektivitas Perangkat Kurikulum Berbasis Cinta Perspektif Hindu Dengan Sistem Rekomendasi Berbasis Deep Learning

Abstract

This study developed and tested a Hindu-perspective Love-Based Curriculum (KBC) learning toolkit, combined with a deep learning-based adaptive recommendation system prototype to enhance students' internalization of affective values ​​and cognitive mastery. The objectives of the study were to design, validate, and test the effectiveness of the toolkit (syllabus, lesson plans, thematic modules, assessment instruments) and to assess the technical performance and pedagogical relevance of the recommendation prototype. The research method combined R&D (the ADDIE framework and Borg and Gall) with a design-based research approach. Evaluation included expert validation, pilot testing, system log analysis, qualitative data (interviews, observations), and a quasi-experimental pretest-posttest effectiveness test (experimental and control groups, n≈30 each). The main results: the toolkit was deemed content valid (mean ICVI = 0.86; SCVI = 0.92) and reliable (Cronbach's α = 0.82); the experimental group demonstrated significant improvements in cognitive and affective outcomes compared to the control group (p < 0.01; Cohen's d ≈ 0.65). The deep learning prototype achieved practical performance (AUC ≈ 0.74; adequate precision/recall), and recommendation relevance positively correlated with improved learning outcomes. Qualitative findings explain the mechanisms of change: structured experiences, cross-perspective value dialogue, and adaptive feedback drive the transfer of values ​​into action. In conclusion, integrating KBC contextualized within Hindu tradition and adaptive technology can serve as an operational model for higher religious education, with the caveat that faculty training, infrastructure improvements, and strengthening data ethics policies are needed for further scale-up.