【作者】 方琦；

【导师】 柳有权；

【作者基本信息】 长安大学 ， 计算机软件与理论， 2018， 硕士

【摘要】 随着人们生活水平的提高,饮食健康成为大众关心的焦点。科学合理的饮食有利于身体健康,同时对疾病的治疗有着非常重要的作用。由于国内营养师匮乏,人们需要采取一种有效方式来进行饮食规划。在这样的问题背景下饮食推荐应运而生,它能辅助用户进行科学的饮食决策。饮食推荐需兼顾兴趣与营养两个方面,如何为用户提供符合营养和兴趣双需求的饮食推荐服务是一个很有实际应用意义的研究课题。本文围绕营养均衡与个性化饮食需求对个性化健康饮食推荐方法进行研究,研究内容如下:1.改进了快速非支配遗传算法。针对饮食营养均衡在多目标优化过程中存在的求解过程复杂、计算速度缓慢、容易陷入局部最优等问题,本文对传统的快速非支配遗传算法进行了改进。该改进算法引入差分变异算子来增强算法的局部搜索能力,通过优化基准的测试函数,表明了该算法能够避免陷入局部最优。针对该算法搜索精度不高的问题,引入了变缩放因子策略,从而更好地兼顾了算法的收敛速度和可靠性。2.改进了协同过滤算法。针对传统的协同过滤算法需解决的用户饮食兴趣问题,本文提出了综合运用协同过滤推荐算法中的Slope One算法、K-Means聚类算法、相似度算法的解决方案。提出了基于K-Means聚类的加权Slope One算法来实现推荐功能。3.提出了一种个性化健康饮食推荐方法。本文通过Scrapy爬虫框架获取菜谱数据与食材营养成分数据,基于改进的NSGA2算法实现营养配餐的优化,基于改进的协同过滤算法实现用户饮食偏好的推荐。并将其应用到饮食推荐中,实现了个性化营养菜品推荐功能。实验结果表明,本文提出的个性化健康饮食推荐方法,能够设计出满足用户喜好和营养双需求的配餐方案,很好地平衡了个性化与营养均衡之间的关系,达到了健康饮食的目的。更多还原

【Abstract】 With the improvement of people’s living standards,dietary health has become the focus of public concern.A scientific and rational diet is beneficial to physical health,and it has a very important role in the treatment of the disease.Due to the lack of domestic nutritionists,people need to take an effective approach to diet planning.In this situation,dietary recommendations came into being.It can assist users in making scientific dietary decisions.Dietary recommendations need to take into account both interest and nutrition.How to provide users with dietary recommendation services that meet the needs of both nutrition and interest is a practical research topic.This article focuses on personalized healthy diet recommendations based on nutritional balance and personalized dietary needs.The research content is as follows:1.Improved fast non-dominating genetic algorithm.In order to solve the problem of complex process,slow calculation speed and easy to fall into local optimum in the process of multi-objective optimization for dietary nutrition balance,this paper proposes an improved fast non-dominating genetic algorithm.Based on NSGA2,this algorithm introduces a differential mutation operator with guidance to enhance the local search ability of the algorithm.By optimizing the benchmark test function,it shows that the algorithm can avoid falling into a local optimum.For the problem that the search accuracy of the algorithm is not high,a variable scaling factor strategy is proposed,so that the convergence speed and reliability of the algorithm are better taken into account.2.Improved collaborative filtering algorithm.In order to solve the user’s dietary interest problem that traditional collaborative filtering algorithm needs to solve,this paper proposes a solution that uses the Slope One algorithm,K-Means clustering algorithm,and similarity algorithm in the collaborative filtering recommendation algorithm.A weighted Slope One algorithm based on K-Means clustering was proposed to implement the recommendation function.3.A personalized healthy diet recommendation method was proposed.The recipe data and ingredients nutrient composition data were obtained through the Scrapy crawler framework.Nutrition catering was optimized based on the improved NSGA2 algorithm,and the user’s diet preference was recommended based on the improved collaborative filtering algorithm.And it was applied to the diet recommendation system to achieve the personalized nutritional dish recommendation function.The experimental results show that the personalized health diet recommendation method proposed in this paper can design a catering program that meets the user’s preference and nutrition needs.It balances the relationship between personalization and nutritional balance well and achieves the purpose of a healthy diet.更多还原