Big Data Framework For Catering Recommendation Systems

With the rapid development of the Internet,big data is extensively used throughout various channels to record and disseminate personal information and user behavior.To identify potential requirements of users,the user data and behavior are collected by exploring different technologies.Mobile catering applications dominate the everyday personal activities of individuals with the speedy development of mobile internet.How to quickly utilize users’ potential needs through user behavior and information is a problem that restaurant businesses in the age of big data must solve is to promote products that may interest users to users.The personalized recommendation accuracy of catering still needs to be improved.The means are still the same,and personalized recommendations do not undergo multi-dimensional data analysis of large-scale data but only rely on users’ browsing information and purchase records to recommend relevant products to users.Such recommendation methods are not efficient for catering recommendations.In the big data era,these methods should be analyzed and deeply excavated from many aspects,to determine the data’s inherent law,the essence,and the data relationships.The development of machine learning presents the opportunity to resolve these problems.In this paper,the recommendation algorithm and its application are studied,and the low-dimensional data features are extracted or combined through machine learning to form high-dimensional features with higher abstraction,and at the same time,they are integrated with the deep learning model WDL.The use of machine learning advantages allows to automatically and effectively learn features,extract essential features,and introduce high-order features into the WDL model.Based on this,a catering recommendation system is developed to provide recommendation services to improve user experience.To achieve the above goals,the main contents of this paper are the following:1.This paper analyzes the recommendation algorithm,its deep learning development and discusses the application and advantages of deep learning in the suggestion field,emphasizing the WDL recommendation model2.In this paper,Word2 vec technology is proposed for mining dishes,completed efficient large-scale mining of text features,and its usage in the recommendation system to extract high-order features and generate candidate sets.3.A Naive Bays classification model of vegetables is proposed.Using machine learning,implicit taste features of high dimensional dishes are extracted.The model provides a hidden automatic extraction feature and relationship learning without manual participation in engineering features.The generated results are used by subsequent recommendation models.4.Generating high-order features through the first two chapters for the WDL model to learn and train.Compared to traditional recommended models,the Experimental statistical results showed that the average recommendation,accuracy,and recall rate of the WDL algorithm increased by 7%,3%,and 2%,respectively.At the same time,the diversity and novelty of the recommended results have been reinforced.5.Based on the above research and AP,the catering recommendation system is designed and implemented,and the above model engineers,and presents the results to users.The system’s function is to provide an effective personalized referral service to users,save users’ time to select goods,improve the user experience,and increase the profits of businesses.