PENGEMBANGAN SAVORYAI: SISTEM REKOMENDASI RESEP MASAKAN BERBASIS BAHAN DAN PREFERENSI KALORI MENGGUNAKAN CONTENT-BASED FILTERING DAN OPENAI API

The high volume of leftover cooking ingredients in households often leads to food waste due to a lack of ideas or references for processing them. This research aims to develop SavoryAI, an intelligent web-based recipe recommendation system that suggests relevan Indonesian dishes based on user-inputted ingredients and calorie preferences. The system integrates Content-Based Filtering (CBF) with cosine similarity and TF-IDF weighting to match user-selected ingredients with recipes in the database. Additionally, OpenAI’s GPT-4o model is utilized to identify food ingredients from uploaded images. The system is implemented using Laravel, Livewire, and TailwindCSS, with data gathered through interviews with household actors and literature reviews. Evaluation was conducted through functional testing (black-box), validity testing, and confusion matrix analysis, using response from household users to determine ground truth. The results show a high accuracy in generating relevant recipe recommendations, with a precision of 1.00, recall of 0.83, and F1-score of 0.91. The results show a high accuracy in generating relevant recipe recommendations. The integration of AI image recognition further enhances usability by enabling automatic ingredient input. The finding highlight the system’s effectiveness in reducing food waste and supporting sustainable cooking practices through personalized recipe suggestions.