Design And Implementation Of The Customer Relationship Management Subsystem In The Centralized Purchasing System

At present,small supermarkets in rural areas generally have problems such as a small quantity of commodities purchased,weak bargaining ability,and lack of funds for large-scale purchases.These problems may lead to dilemmas such as high unit prices and untimely supply,which restricts the further development of these small supermarkets.Therefore,how to use the limited funds to balance the supply and demand between small supermarkets and farmers has become an urgent problem to be solved.This system is based on the key research and development plan of the Ministry of Science and Technology-Green and Livable Villages and Towns Technology Innovation Project(2019YFD1101104),and it takes the cooperative small supermarkets of the Supply and Marketing Cooperative in Shiyan of Hubei province as its research background,and the customer-centric marketing model as its core idea.It is designed and developed based on the actual needs of the small supermarkets.The system mainly includes management of customers' information,prediction of commodity sales,commodity association mining,management of serving members,evaluations of supplier's supply capacity and other significant modules.This dissertation mainly solves two problems,the first problem is the prediction of commodity sales,and the second problem is commodity association mining.For the first one,this dissertation uses the nonlinear least square method to estimate the parameters of the improved Bass diffusion prediction model,which combines the sentiment's analysis of product's reviews and the standard Bass prediction model to predict the sales volume of products.For the second problem,this dissertation uses the H-Apriori(Hash-Apriori)algorithm to find out relationships between commodity.When the system is faced with massive amounts of data,the distributed method of Spark is used to improve the operating efficiency of the algorithm.This system uses B / S architecture and Java language,SSM framework in rear-end takes charge of page Jump,business logic and data operation.The front-end uses Boot Strap framework and j Query to display data and page logic,and the front-end realizes data visualization through ECharts plugin.The database is the relational database,My SQL.The experiment shows that the mean square error of Bass emotional model is 38.53%higher than that of Bass standard model.The H-Apriori algorithm on a Spark cluster with three Worker nodes has a speedup ratio of 2.61 relative to a single machine.With the analysis and management of data for customers' information,the system realizes the prediction for commodity sales,and the mining of commodity groups which have strong associations.This system is extremely practical and convenient for the mall administrator to carry out macrocontrol.