Research And Application Of Quality Food Recommendation System Based On Deep Learning

Since there are tens of thousands of ingredients in real life,according to different cooking styles,different races,cultures and personal tastes,tens of thousands of foods can be made.Deciding what to eat becomes a problem at the right time.Choosing a delicious dish seems to be a difficult task.The explosive growth of food has brought two problems:on the one hand,simply search engines cannot provide targeted personalized services,and it is difficult for diners to obtain food that they do not understand but may be interested in from a large amount of food information;On the other hand,as food producers and producers,how to effectively use information to push to potential target customers has become a problem.This article selects the recommendation system to effectively solve these two problems.However,the application scenario of the recommendation system is far more than the original one,and the collaborative filtering algorithm gradually exposes many deficiencies,including the sparseness,scalability,real-time nature,collaborative interest drift,cold start,etc.of the collaborative filtering algorithm problem.In response to these problems,on the premise of guaranteeing the quality of recommendations,the problems of sparseness,user interest drift,cold start,etc.were studied,and the knowledge of deep learning was introduced.A hybrid recommendation based on convolutional neural network and collaborative filtering was proposed.System,the main work of the paper is as follows:1.Use the Skip-Gram model in Word2Vec to process food information,generate word vectors,and meet the requirements of convolutional neural network input data.Construct a convolutional neural network model to implement a convolutional neural network improved on the Yoon Kim network model.Experiments show that the improved convolutional neural network is superior to the original model in the classification of food data in this paper.2.Combine the user-based collaborative filtering algorithm with the convolutional neural network,and use the improved Yoon Kim network model to extract the category of dishes from the recipe,so that the food recommendation system can be based on the user's vague eating habits in accordance with the user,The information accurately finds and pushes the food that the user may be interested in from the massive food and catering information.To a certain extent,the problem of user cold start has been improved.