Personalized Menu: a New Contextual Collaborative Recommender System

The varied number of suggested meals put people who spend most of their time outside to a confusing situation. In addition, the proposed items (meals) are standard for all people, meaning that they receive the same menu, although they differ in many attributes such as age, weight, health status, eating habits and tastes. A recommendation based on nutritional value can help people to prevent or control certain diseases such as diabetes, hypertension and obesity. However, learning about people's food preferences and making recommendations that simultaneously appeal to their health expectations, which can be very difficult, is really a very difficult task. Many existing recommendation systems do not take into account the context in which the user is placed and often suffer from the cold start system problem. Through this work, we propose a new collaborative and contextual recommendation system that allows the target user to be directed towards healthy menu of tasty meals. For this, the proposed approach was named Healthy and Tasty (H&T). H&T is a recommendation system based on a multilayer architecture, where the highest layer is based on the pre-filtering algorithm, which eliminates the cold start system problem, and is based mainly on the thresholds of the compatible nutritional elements with the user profile, thus ensuring the healthy factor in the recommendations. The middle layer based on collaborative filtering algorithm works in particular to ensure the taste factor, while the deepest layer is the pertinent layer based on the post-filtering algorithm, where the recommendation will be much more adapted to the environment of the user. The proposed approach has been realized via an Android application where the experimental results are promoters.