基于小样本学习的公路物流车主接单预测模型  

Prediction Model of Car Owner Taking Orders Based on Few-shot Learning

作　　者：达明添 袁凯 王东 孙知信[1] DA Ming-tian;YUAN Kai;WANG Dong;SUN Zhi-xin(Jiangsu Provincial Postal Big Data Technology and Application Engineering Research Center,Nanjing University of Posts and Telecommunications,Nanjing 210003;Anhui Yougu Express Intelligent Technology Co.,Ltd.,Wuhu 241300;Jiangsu Manyun Software Technology Co.,Ltd.,Nanjing 210003,China)

机构地区：[1]南京邮电大学江苏省邮政大数据技术与应用工程研究中心,江苏南京210003 [2]安徽邮谷快递智能科技有限公司,安徽芜湖241300 [3]江苏满运软件科技有限公司,江苏南京210003

出　　处：《物流工程与管理》2024年第12期32-36,共5页Logistics Engineering and Management

基　　金：贵州省科技支撑项目([2023]一般272)。

摘　　要：随着电商平台的崛起,物流行业迅速发展,车货平台应运而生。尽管车货平台迅猛发展,但也面临一些挑战。不同的车主对订单有不同的要求,如果车货平台无法了解车主的接单要求,就无法实现精确的车货匹配。尽管传统的深度学习方法可以进行车主接单预测,但它们需要大量标记数据进行模型训练,而且数据收集和人工标记过程耗时费力。为解决这一问题,文中提出了一种基于SAE-SCNN模型的车主接单预测模型,该模型结合了小样本学习和车货匹配的特点。SAE-SCNN模型由堆栈式自编码器(SAE)和孪生卷积神经网络(SCNN)组成,与传统方法相比,该方法只需要少量标记数据样本即可进行模型训练。通过对某车货匹配平台的数据集进行对比实验,结果表明,文中提出的SAE-SCNN模型在车主接单预测方面展现出良好的性能。With the rise of e-commerce platforms and the rapid development of the logistics industry,vehicle and freight platforms have emerged.Despite the rapid development of car and freight platforms,they also face some challenges.Different car owners have different requirements for orders,and if the car and freight platform cannot understand the car owner's order receiving requirements,accurate car and freight matching cannot be achieved.Although traditional deep learning methods can be used for predicting car owner's orders,they require a large amount of labeled data for model training,and the process of data collection and manual labeling is time-consuming and laborious.To address this issue,this paper proposes a car owner order prediction model based on the SAE-SCNN model,which combines the characteristics of Few-shot Learning and car and freight matching.The SAE-SCNN model consists of a Stack Auto-encoders(SAE)and a Siamese Convolution Neural Network(SCNN).Compared with traditional methods,this method only requires a small number of labeled data samples for model training.Through comparative experiments on a dataset of a certain car and freight matching platform,the results show that the SAE-SCNN model proposed in this paper exhibits good performance in predicting vehicle owner order acceptance.

关 键 词：车货匹配 小样本学习 接单预测 堆栈自编码器 孪生网络 

分 类 号：U492.3[交通运输工程—交通运输规划与管理]