Design And Implementation Of C2B Used Car Merchant Bidding Recommendation System Based On Elastic Stack

In recent years,with the growth of domestic car ownership year-on-year and the expansion of the used car transaction market,used car transactions have gradually shifted from the traditional offline model to the online platform.Among them,the C2 B model relies on the advantages of fast transactions,short transaction cycles,and stable buyer resources to occupy a certain market share.Merchants play a key role as the downstream transaction in the C2 B model.In this era of vehicle source information explosion,there are many vehicles in each auction.How to quickly and effectively recommend the preferred vehicles for merchants,increase the number and efficiency of effective quotations,improve merchants’ satisfaction,and reduce the rate of unsold vehicle.It is one of the core objectives of the C2 B used car trading platform.Based on the actual business scenario of the used car trading platform K,combined with the characteristics of the industry,users and used cars,this thesis designs and implements a C2 B used car merchant bidding recommendation system based on Elastic Stack.The back end of the system is developed based on the Django framework of MTV mode,and data feature projects are completed by big data modules such as sklearn.The data of system is saved by My SQL,Mongo DB,and Redis according to the data characteristics,and the front end uses the VUE framework to realize page display and rendering.The main research work of this thesis can be divided into the following points:First,for the problem of used car “one car one status” and high timeliness of bidding vehicles in the C2 B mode,this thesis proposes a vehicle popularity scoring algorithm based on the weight of click-through rate and effective bid rate in the offline recommendation module,thereby completing the Popular recommendation;At the same time,the built merchant preference model is used to realize content-based recommendation.The combination of the two algorithm solves the problem of cold start of users and vehicles.Secondly,aiming at the problem of auction vehicles,this thesis proposes an unsold weighting algorithm based on the number of valid bids and auction time in the recommendation engine module.The algorithm can improve the exposure of vehicles with high possibility of passing auction according to the current auction time,and effectively reduce the passing rate of auction vehicles and the cost loss caused by passing auction.Through online testing,the rate of unsold vehicles can be reduced by1.85%,and the algorithm has a certain versatility for improving the exposure of longtail items in the auction scene.Third,build a set of highly available log collection,filtering and analysis modules.The Elastic Stack is used to realize the asynchronous collection,filtering and indexing the logs of the merchant’s click and bid behavior,and then the recommendation system extracts and statistically analyzes the logs through the ES REST interface.Fourth,design and implement a bidding recommendation system based on the above algorithm.Merchants can use this system to quickly locate their preferred vehicle in the auction and complete the bidding process,which improves the efficiency of their browsing and quotation.