Research On Personalized Recommendation Methods Of Agricultural Products E-commerce Based On User Portraits

With the development of China's major agricultural product e-commerce platforms,a large number of users and agricultural product information have been generated,and the phenomenon of “information overload” has appeared,making it more difficult for agricultural users to browse the products they are interested in,and they cannot purchase them in a timely and effective manner.Agricultural products that meet their own needs.Therefore,further mining is carried out on consumer behavior and other information generated when consumers interact with e-commerce platforms,and group dynamic user portraits are constructed to predict users interests and preferences,and personalized recommendation systems are used to solve the “information overload”problem.In this thicsis,the research on agricultural product user portraits mainly includes three parts: basic information sub-portrait,behavior sub-portrait and agricultural product domain knowledge sub-portrait.It studies the update technology of user portraits,and builds clusters based on the agricultural product domain knowledge map by improving a kernel fuzzy clustering algorithm.Group dynamic user portraits,and further combine dynamic portrait technology to study the personalized recommendation methods of agricultural products e-commerce,and finally push the recommended agricultural products to consumers through the e-commerce platform system.The main research contents are as follows:(1)Research on group dynamic user portraits.From the perspective of user needs,this thicsis establishes a knowledge map in the field of agricultural products,analyzes the historical information generated by consumers on the e-commerce platform,and combines data mining and other means to generate user tags,and further studies the combination of time series and user portraits to achieve dynamics Update user portraits.A fuzzy kernel C-means algorithm(FKCM)based on kernel function is improved by adopting the nearest neighbor propagation(affinity propagation,referred to as AP)algorithm to cluster user portraits.Based on the above research,a group dynamic user portrait model based on the knowledge map of agricultural products is constructed.In order to verify the validity and feasibility of the model,the group user portraits obtained by traditional algorithms are compared,and the user portraits are designed for performance experiments on recommendation.Experiments show that the group user profile constructed in this thcsis has improved and improved in accuracy and recall.(2)Research on the methods of personalized recommendation of agricultural products.Based on the data in this thcsis in the calculation of the collaborative filtering recommendation algorithm(CF),in view of the consumer scoring habits and other reasons,resulting in large calculation errors and other problems,after analysis,the improved cosine similarity calculation formula is used to calculate the user based on the above similarity calculation method,the weighted scoring method is used for collaborative filtering recommendation,and combined with the user portrait technology,the user historical data is analyzed,and the user interest tag information is extracted from it,which improves a dynamic user portrait and collaboration Filtering(DUCF)fusion recommendation algorithm,design experiment and traditional recommendation algorithm(CF,UCF)comparison,the overall performance has been improved.(3)The design and implementation of a personalized recommendation system for agricultural products based on the dynamic user portrait and collaborative filtering(DUCF)recommendation algorithm.Through the functional analysis and database design of the system,and based on the Django framework,the DUCF-based personalized recommendation function module for agricultural products is designed and developed.At the same time,the recommendation system in this thcsis realizes the complete functions of browsing,collecting,shopping,adding shopping carts,etc.for users Provides a smooth shopping experience.