RESEARCH AND IMPLEMENTATION OF COLD START DIET RECOMMENDATION ALGORITHM BASED ON MULTI-MODAL
In order to help users find recipes in line with their personal dietary preferences from a large number of recipes,a cold start algorithm for dietary preference recommendation based on multi-modal is proposed.The algorithm was divided into online and offline phases.In the offline stage,the semantic vector of each dimension of recipe information was obtained through representation learning.In the online stage,based on the idea of iteration and graph similarity propagation,the user's preference for all recipes was obtained and recommended through a small amount of preference feedback of recipes.The experimental results show that the accuracy of the cold start diet recommendation algorithm based on multi-modal achieves the level of practical engineering application.
国家科技期刊平台
NSTL
万方数据
维普
