Sistem Rekomendasi Resep Makanan Penutup dengan Input Bahan yang Tersedia Menggunakan Metode Content-Based Filtering

Kepercayaan diri merupakan salah satu aspek penting dalam perkembangan remaja, namun fenomena ketidakpercayaan diri (insecurity) akibat berbagai faktor, seperti kesenjangan sosial ekonomi dan tekanan akademik, masih sering ditemukan di lingkungan sekolah. Penelitian ini bertujuan untuk mendeskripsikan tingkat kepercayaan diri peserta didik sebelum dan sesudah diberikan layanan bimbingan kelompok berbasis Cinema Therapy, serta menganalisis perubahan kepercayaan diri setelah intervensi pada peserta didik di SMA Negeri 1 Palopo. Penelitian ini menggunakan pendekatan kuantitatif dengan desain quasi-experiment tipe One Group Pre-test-Post-test. Sampel penelitian terdiri atas sembilan peserta didik kelas X MIPA 1 yang teridentifikasi memiliki tingkat kepercayaan diri rendah berdasarkan hasil pre-test, dengan teknik purposive sampling. Data dikumpulkan menggunakan angket skala kepercayaan diri model Likert dan dianalisis menggunakan Paired Sample t-Test. Hasil penelitian menunjukkan bahwa rata-rata skor kepercayaan diri meningkat dari 58,33 pada pre-test menjadi 65,44 pada post-test. Hasil uji hipotesis menunjukkan nilai signifikansi sebesar 0,004 (p < 0,05), yang mengindikasikan adanya perbedaan skor kepercayaan diri sebelum dan sesudah intervensi. Meskipun demikian, mengingat penelitian menggunakan desain tanpa kelompok kontrol dan melibatkan jumlah sampel yang relatif kecil, hasil penelitian perlu diinterpretasikan secara hati-hati. Temuan ini menunjukkan bahwa layanan bimbingan kelompok berbasis Cinema Therapy berpotensi meningkatkan kepercayaan diri peserta didik dalam konteks kelompok kecil yang diteliti, melalui proses refleksi diri, pengalaman vikarius dari media film, dan diskusi kelompok yang mendukung pengembangan keyakinan diri.

Dessert Recipe Recommendation System Based on Available Ingredients Using The Content-Based Filtering Method

Abstract

The availability of various digital recipe sources does not fully assist users who want to determine dessert recipes based on the ingredients available at home, as recipe searches are generally conducted by searching for specific food names or types. This study develops a web-based dessert recipe recommendation system that not only ranks recipes based on ingredient similarity but also displays the percentage of ingredient matches and the ingredients that are still required by users. The system was developed using the Content-Based Filtering method with the Term Frequency-Inverse Document Frequency (TF-IDF) algorithm and Cosine Similarity. The dataset consisted of 150 dessert recipes, while the evaluation used 20 ingredient queries, resulting in 3,000 query-recipe pairs. Recommendation relevance was determined using a rule-based approach, with a minimum threshold of 50% of the user's input ingredients being present in the recipe. System performance was evaluated using a Confusion Matrix with Accuracy, Precision, Recall, F1-Score, and Error Rate as evaluation metrics. The results showed an Accuracy of 86.8%, Precision of 70.7%, Recall of 67.7%, F1-Score of 69.2%, and Error Rate of 13.2%. These results indicate that the system is capable of generating recommendations based on ingredient compatibility with measurable performance. The contribution of this study lies in developing a recipe-matching mechanism that combines TF-IDF and Cosine Similarity-based similarity with information on the proportion of available ingredients and the ingredients that are still required. However, the interpretation of the results should consider that the relevance labels were determined using a rule-based approach and were not independently established through expert or user assessment.