Personalized Healthy Diet Recommendation Service Research

Tourism is composed by food, accommodation, transportation, travel, purchase and entertainment, from which we can see that the diet occupies the important position in the tourism services. At the same time, the scientific and reasonable diet is not only beneficial to keep fit, but also plays a very important role in the treatment of the disease. In today's society, it has become a very popular topic. Limited knowledge of health sciences, people incline to help themselves plan healthy diets through the Internet. The Internet information has been greatly increased and gradually outstripped the range which individual could accept, process and use effectively, as a result, people begin to get lost in thousands of diet information. Diet recommendation can help users find the content they want to look for from the massive diet data. Diet recommendation need to comprehensively consider two aspects of personalization and health. The key to it is to understand types of users'diet and appropriate intakes, which will help users avoid uneven or improper diet, increase their life quality and reduce health care costs. It has become a subject needs to be solved about how to provide users with a series of personalized healthy diet recommendation service which can both well satisfy their preferences and conform to the standards of nutrition and health.Healthy diet recommendation service is a nonlinear multi-objective optimization problems. The intakes of nutrients and energy both need to be optimized. Personalized diet recommendation service can be achieved by mining user generated content of food interactive community, using LDA topic model to analyze the text information posted by users, mining user dietary interest and eating habits. In the process, it involves many aspects of data and knowledge, including users, food, nutrition and health. So this thesis uses ontology for the relevant knowledge modeling, storage and matching to solve the information sharing and interaction to in semantic level, and analyzes the framework of diet recommendation service with the description logic reasoning of diet ontology. We research personalized healthy diet recommendation service base on ontology, multi-objective genetic algorithm and LDA topic model. The main work is as follows:(1) Built the diet ontology, propose the framework of personalized healthy diet recommendation service based on diet ontology. Diet ontology is built based on the general user model ontology, common food recipes and food composition data, theory of nutrition, which contains three parts:the user model, food model and nutrition model. The proposed framework of personalized healthy diet recommendation service based on diet ontology is able to establish a link among users, diets and disease, achieve the user information understanding and dietary knowledge sharing, and provide a knowledge base for the personalized healthy diet recommendation service.(2) A multi-objective genetic algorithm based on the dietary records is proposed by adding long-term and cumulative effects of diets to the recommendation strategies. This thesis analyzes user information based on diet ontology, establishes multi-objective diet recommendation model, and recommends food and food consumption for users by using multi-objective genetic algorithm based on the dietary records. The qualified rate of recommendations can demonstrate the feasibility and effectiveness of multi-objective genetic algorithm based on the dietary records. Comparing with the traditional genetic algorithm based on the random weighting, the cumulative error rate of the proposed algorithm closer to zero. In a healthy perspective, it verifies that the proposed algorithm is more suitably for personalized healthy diet recommendation service.(3) Based on the LDA model, we put forward a dishes recommendation method based on user dietary interest topic model. In this paper, user generated content of food interactive community is collected and stored based on diet ontology to enrich the recommendation service knowledge base based on diet ontology. We analyze user generated content of food interactive community, determine the number of topics according to the perplexity and build the user dietary interest model. Based on this model, dishes, which are associated with the recommended food in the previous section, are recommended. This method can help the users select dishes more in line with the user preferences. The Experiments, conducted with the real data sets, show that the user dietary interest topic model based on LDA can be applied to personalized dishes recommendation.