Research On Intelligent Recommendation Model Of Transport Capacity Resources Of NTOCC Platform

Since its birth in 2013,Non-Truck Operating Common Carrier(NTOCC)platform has attracted the attention on the industry and academia by virtue of its unique "Internet+ logistics" mode.It solves the problem of information asymmetry in the traditional mode of finding carrier and solves the problems of high logistics cost and low logistics efficiency in the market of transportation capacity resources trading in China to some extent.With the development of NTOCC platform and the increase of user scale,the explosive growth of intelligent logistics puts forward requirements for automation and refinement of logistics links.The current platform's capacity resource recommendation mechanism cannot meet the needs of both the carrier and the shipper.The inefficient matching mode leads to the carrier's high no-load rate and the income level is not significantly improved,It also brings higher logistics cost and lower efficiency to shipper.Therefore,how to improve the strategy of transportation resources recommendation to reduce cost and efficiency of both freight transport parties becomes a topic worthy of study.In view of the above problems,this paper comprehensively considers the influence of vehicle utilization rate and user correlation on the recommendation model,and proposes an intelligent recommendation strategy for transportation resources of NTOCC platform,so as to achieve cost reduction and efficiency improvement of both the carrier and the shipper.The main research contents and research contributions of this article are as follows.First of all,through field investigation,literature induction and figuring out NTOCC platform problems existing in the current recommendation mechanism,and based on this platform user attributes analysis,building user efficiency index system,with the meaning of performance indicators and specific quantitative method is given;Secondly,this paper introduces a new quantitative method: using the idea of collaborative filtering to calculate the similarity between the owner and the owner,as well as the similarity between the owner and the owner,In order to calculate the user correlation,a scheme of post fusion of graph model and collaborative filtering model is proposed;In the training stage of machine learning model,according to the shortcoming that LR logistic regression model cannot deal with continuous features well,a GBDT + LR fusion scheme is proposed,and the evaluation index of the recommended algorithm is given;At last,some data of a NTOCC platform is taken to build an example to further verify the validity and stability of the model,and management explanation is given according to the prediction results.