<![CDATA[Differences in student learning styles demand adaptive and personalized learning strategies. This study aims to develop a web-based learning strategy recommendation system utilizing Artificial Neural Network (ANN) to model the relationship between students' learning styles and appropriate learning strategies. Learning style identification was conducted using the Felder-Silverman questionnaire encompassing four dimensions: active–reflective, sensing–intuitive, visual–verbal, and sequential–global. The study employed a Research and Development (R&D) method with the 4D model and Personal Extreme Programming (PXP) approach. Data were collected from 25 seventh-grade students at SMP Negeri 1 Tomilito. A Multilayer Perceptron ANN model was trained using the backpropagation algorithm over 3,000 epochs, yielding a Mean Squared Error (MSE) value of 0.0541, indicating a relatively low prediction error rate. System feasibility testing obtained a score of 85.42%, categorized as "Very Feasible." The developed system is capable of identifying students' learning styles and automatically generating learning strategy recommendations, thereby potentially supporting teachers in designing more adaptive and personalized learning experiences. Perbedaan gaya belajar siswa menuntut adanya strategi pembelajaran yang adaptif dan terpersonalisasi. Penelitian ini bertujuan mengembangkan sistem rekomendasi strategi pembelajaran berbasis web yang memanfaatkan Artificial Neural Network (ANN) untuk memodelkan hubungan antara gaya belajar dan strategi pembelajaran yang sesuai. Identifikasi gaya belajar dilakukan menggunakan kuesioner Felder-Silverman yang mencakup empat dimensi: aktif–reflektif, sensori–intuitif, visual–verbal, dan sequential–global. Penelitian menggunakan metode Research and Development (R&D) dengan model 4D dan pendekatan Personal Extreme Programming (PXP). Data dikumpulkan dari 25 siswa kelas VII SMP Negeri 1 Tomilito. Model ANN Multilayer Perceptron dilatih menggunakan algoritma backpropagation dengan 3000 epoch dan menghasilkan nilai Mean Squared Error (MSE) sebesar 0,0541, yang mengindikasikan tingkat kesalahan prediksi yang relatif rendah. Hasil uji kelayakan sistem memperoleh skor 85,42% dengan kategori "Sangat Layak". Sistem yang dikembangkan mampu mengidentifikasi gaya belajar siswa dan memberikan rekomendasi strategi pembelajaran secara otomatis, sehingga berpotensi mendukung guru dalam merancang pembelajaran yang lebih adaptif dan personal.]]>
