基于深度学习的物流服务交易群智推荐算法的研究

Research on Deep Learning-Based Recommendation Algorithm for Group Wisdom of Logistics Service Transactions

针对物流服务交易中物流服务商和客户供需双方之间的交易信息匹配精确度不高、推荐效率低的问题,提出了一种基于深度学习的物流服务交易群智推荐算法。首先,对传统物流服务交易流程和影响双方交易信息匹配精度不高的因素进行分析,根据多目标优化与相似性理论,构建供需双方物流交易信息匹配模型。其次,利用多目标粒子群算法,建立物流服务交易信息群智匹配算法。再次,运用深度神经网络的分类优势,对客户的喜好、服务商选择、交易匹配信息进行学习和训练,建立DeepFM的训练模型,生成优化的物流交易分类信息,提出基于深度学习的物流服务交易群智推荐算法。最后,在互联网物流服务交易环境下对算法进行验证匹配精度为68%,推荐效果为76%。实验结果表明,该算法是一种快速匹配,精确度和推荐成功率高的群智物流服务交易算法。

Aiming at the problems of low accuracy of matching transaction information between the supply and demand sides of lo⁃gistics service providers and customers in logistics service transactions and low recommendation efficiency,a deep learning-based rec⁃ommendation algorithm for logistics service transactions with group wisdom is proposed.Firstly,the traditional logistics service transac⁃tion process and the factors affecting the low accuracy of transaction information matching between the two parties are analyzed,and the logistics transaction information matching model between the supply and demand sides is constructed according to the multi-objective optimization and similarity theory.Secondly,the multi-objective particle swarm algorithm is used to establish the logistics service trans⁃action information swarm wise matching algorithm.Again,using the classification advantage of deep neural network,we learn and train the customer’s preference,service provider selection and transaction matching information,establish the training model of DeepFM,generate the optimized logistics transaction classification information,and propose the group wise recommendation algorithm of logis⁃tics service transaction based on deep learning.Finally,the algorithm is verified in the Internet logistics service transaction environ⁃ment with matching accuracy of 68%and recommendation effect of 76%.The experimental results show that the algorithm is a group wise logistics service transaction algorithm with fast matching,high accuracy and recommendation success rate.