Recommendation of Food Items for Thyroid Patients Using Content-Based KNN Method

Abstract

AbstractFood recommendation system has become a recent topic of research due to increase use of web services. A balanced food intake is significant to maintain individual’s physical health. Due to unhealthy eating patterns, it results in various diseases like diabetes, thyroid disorder, and even cancer. The choice of food items with proper nutritional values depends on individual’s health conditions and food preferences. Therefore, personalized food recommendations are provided based on personal requirements. People can easily access a huge amount of food details from online sources like healthcare forums, dietitian blogs, and social media websites. Personal food preferences, health conditions, and reviews or ratings of food items are required to recommend diet for thyroid patients. We propose a unified food recommendation framework to identify food items by incorporating various content-based features. The framework uses the domain knowledge to build the private model to analyze unique food characteristics. The proposed recommender model generates diet recommendation list for thyroid patients using food items rating patterns and similarity scores. The experimental setup validated the proposed food recommender system with various evaluation criteria, and the proposed framework provides better results than conventional food recommender systems.KeywordsContent-basedK-Nearest neighborTerm frequencyInverse document frequencyFood preference matrix