This paper details the development and empirical evaluation of an adaptive learning system aimed at enhancing digital science literacy among junior high school students in Indonesia. The primary challenge addressed is the limitation of one-size-fits-all educational models. Our research proposes a user-centric solution, the Sistem Rekomendasi Cerdas (SRC), developed through a co-design methodology. The system’s core is a User-based Collaborative Filtering (UBCF) algorithm. Its effectiveness was evaluated through a pre-test/post-test experimental study involving 60 students, divided into an experimental and a control group. Quantitative results show that the experimental group achieved a significantly higher increase in science literacy scores (p < 0.001) compared to the control group. Qualitative findings from interviews with the experimental group reveal that the platform enhanced learning motivation, content relevance, and helped overcome learning barriers. This study concludes that the SRC, developed via a co-design model, is a highly effective tool for improving digital science literacy, demonstrating that a user-centered approach is fundamental to creating impactful educational technology.
 
Keywords - Adaptive Learning, Co-Design, Recommender System, Digital Literacy, Usability.