Data-Driven Fish Recommendation System Based on Taste Preferences

This paper proposed a machine learning-based fish recommender system that is personalized to fish types related to taste. Through the k-Nearest Neighbor (k-NN) with Cosine Similarity and a threshold-based recommender, the system predicts fish types based on attributes of sweetness, saltiness, tenderness, oiliness, fishiness, texture, acidity, and metallic. The system does so without the customer tasting the fish, therefore, improving their purchasing experience by making thoughtful sourcing decisions. The developed recommendation system is suitable for seafood and even food purchased online. The recommender system allows users to select food items based on their taste preferences. Users receive a thoughtful and accurate recommendation which therefore results in greater user satisfaction, less waste and makes the purchase of food products online more efficient. This machine learning approach is a breakthrough in learned e-commerce because it changes the way users purchase fresh food items online.