Research On Personalized Dietary Recommendation Services Based On Dietary Behavior

With the development of social economy and the improvement of human life, people pay attention to the problem of diet and health increasingly, and healthy diet is getting more and more popular, a series of diseases caused by improper diet, such as diabetes, high blood pressure, obesity, and so on, have long been served as the wake-up call to remind people to keep balance diet. As the general public, it is difficult for them to judge whether their three meals in a day is scientific, reasonable or balanced. In order to solve this problem, the dietary recommendation system realized by using the computer technology arises at the historic moment.At present, there are many nutritional catering software, most of which can analyze the users' nutritional composition according to the users' input information (age, height, weight, labor intensity), then give the nutritional advice, they will automatically match the nutritional needs of food combination from the massive food bank and to choose for people according to their own preferences. In addition, recipe sites have been emerged one after another and can generate several sets of recipes for users automatically according to the seasons, food category, meal times and other certain conditions. Although these patterns can solve the problem of personal dietary "difficult to choose" in some extent, they cannot balance the relationships between the personal taste preference needs and nutritional needs. So it is difficult to ensure the quality of catering results. Therefore this paper puts forward a research method of personalized diet recommendation service, mainly to design recipes scheme which can meet the taste and nutritional needs of the users, the main work is as follows:1. Healthy diet analysis:thinking about many people have unhealthy eating problems, this paper mainly analyzes students' scientific and proper diet of students from three aspects:the diet variety, rationality of diet time and balanced nutrition by taking the campus card and students' dietary record data for example, and provides the referential suggestion;2. Research on personalized nutritional recipes recommendation based on multi-objective optimization:thinking about the actual demand of essential nutrients for human body, this paper puts forward a dynamic tuning nutrition model to calculate the users' actual demands of various nutrients daily, further, the multi-objective particle swarm optimization makes diet of three meals in a day recommended for the users, which can meet the users' actual needs of various nutrients as possible as they can, so as to realize the personalized nutritional recipes recommendation based on multi-objective optimization;3. In order to better coordinate the relationship between food nutrition and taste, this paper introduces the concept of the nutritional recipes preference degree and its calculation method, and uses the average times of the food that users eat (hereinafter referred to as frequency) to indicate the users' preferences. Firstly, the food which the users like and the forecast frequency of the food is obtained by using collaborative filtering combined with multi-objective optimization methods. The algorithm can not only use multi-objective optimization algorithm for grouping users, making each user in the target user group be as much as possible to improve the target user all kinds of nutrition elements of unreasonable intake, but also enable the collaborative filtering algorithm to generate the food recommendation of the target users' preferences on the basis of user groups. Then through computing the total frequency of foods which the user likes in the nutritional recipes, we get the preference degree of the recipes and recommend the recipes with higher preference degree.