Abstract: Food culture is an important part of Chinese civilization. The exploration of eight cuisines in food culture is conducive to the inheritance and promotion of Chinese civilization. A visual analysis system DietVis is proposed in order to make reasonable diet recommendations to people. Firstly, a novel view is designed to express the multidimensional attributes of cuisines to help users analyze the constituent features of cuisines. Secondly, through clustering analysis of ingredients and cooking techniques of recipes, users can explore the association between recipes from multiple perspectives. Thirdly, combine the association rules of recipe collocation to recommend to meet the retrieval needs of different users. Finally, a complete visual analysis system of dietary data is constructed through multi-view linkage, which supports users to explore and compare cuisines and recipes in multiple layers. Through case analysis and user evaluation experiment, and using Likert scale to analyze the questionnaire, the effectiveness of visual view and the practicability of the system are verified.