Research On Recommendation Of Agricultural Products Based On Conditional Generative Adversarial Net

By analyzing the historical interaction between users and items,the recommendation system obtains users’ preferences,excavates users’ potential needs,and then recommends products that users may be interested in.In the current common agricultural product recommendation system,most models do not make full use of the data such as comments and scores of users interacting with agricultural products,so it is difficult to learn more accurate user preferences,resulting in poor recommendation effect.Conditional generative adversarial net is based on the generative adversarial net,which can better fit the distribution of input data.In order to learn more accurate user preferences,by fusing BERT and graph convolution neural network,this dissertation extracts user features and agricultural product features from the interactive comment information between users and agricultural products,and then combines the conditional generative adversarial net for recommendation,so as to improve the effect of agricultural product recommendation.Based on the systematic analysis of the research on agricultural products recommendation model and generative adversarial net,aiming at the problem that the current agricultural products recommendation model does not make full use of comment data and is difficult to accurately learn user preferences,this dissertation proposes a recommendation of agricultural products based on conditional generative adversarial net(RACGAN).The main research of this dissertation is as follows:(1)Feature extraction of users and agricultural products by integrating BERT and graph convolution neural network technology.In view of the problems that the current agricultural product recommendation model fails to make full use of the comment information of users and agricultural products and learn user preferences better,this dissertation integrates BERT and graph convolution neural network to make better use of the comment information and extract higher quality features of users and agricultural products.(2)This dissertation constructs a recommendation model of agricultural products based on conditional generative adversarial net.In order to better learn users’ preferences and improve the recommendation effect,this dissertation constructs an agricultural product recommendation model RACGAN based on conditional generative adversarial net,which is composed of a generator and a discriminator.The generator takes user characteristics and agricultural product characteristics as input,and the goal is to predict the probability of interaction between users and agricultural products.The discriminator takes the user characteristics and interaction matrix as the input,and the goal is to judge whether the input interaction matrix is real or predicted by the generator.Finally,some agricultural products with the highest interaction probability predicted by the generator are selected for recommendation.(3)Through relevant experiments on real agricultural product data sets,this dissertation explores the impact of different super parameters on the effect of agricultural product recommendation,and verifies the effectiveness of RACGAN proposed in this dissertation.At the same time,RACGAN is compared with other current main recommendation models on agricultural product data sets.The experimental results show that the RACGAN proposed in this dissertation has certain advantages over other recommendation models in recommendation effect.