主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2024, Vol. 32 ›› Issue (11): 270-278.doi: 10.16381/j.cnki.issn1003-207x.2022.0102

• • 上一篇    下一篇

基于超图嵌入表示的商品分仓模型

刘顺1, 黄晓宇1(), 鲜征征2, 左文明1   

  1. 1. 华南理工大学电子商务系,广东 广州 510006
    2. 广东金融学院互联网金融与信息工程学院,广东 广州 510520
  • 收稿日期:2022-01-14 修回日期:2022-06-15 出版日期:2024-11-25 发布日期:2024-12-09
  • 通讯作者: 黄晓宇
  • 基金资助:
    广东省哲学社会科学规划项目(GD21CGL02); 人工智能与数字经济广东省实验室(广州)青年学者项目(PZL2021KF0027); 广东省自然科学基金项目(2020A1515010830); 华南理工大学中央高校基本科研业务费项目(XYMS202107); 广东省普通高校人文社会科学研究重点项目(2018WZDXM032); 广州市科技计划项目(202002030473); 广东省普通高校创新团队项目(2021KCXTD079)

A Commodity Warehouse Allocation Model Based on Hypergraph Embedding

Shun Liu1, Xiaoyu Huang1(), Zhengzheng Xian2, Wenming Zuo1   

  1. 1. Department of Electronic Business,South China University of Technology,Guangzhou 510006,China
    2. School of Internet Finance and Information Engineering,Guangdong University of Finance,Guangzhou 510520,China
  • Received:2022-01-14 Revised:2022-06-15 Online:2024-11-25 Published:2024-12-09
  • Contact: Xiaoyu Huang

摘要:

电子商务的高速发展要求电商企业能快速地响应消费者的需求。当前,越来越多的电商企业通过构建分布式的配送中心,使库存尽可能地靠近消费者,从而实现分区域的就近发货和就近配送。然而,在多仓分布的设置下,不合理的商品分仓将导致大量的订单被拆分配送,这不仅会给消费者带来不便,还将使电商企业付出额外的运营成本。本文提出了一个以最小化被拆分订单的总量为目标的商品分仓优化模型。首先,基于已有的历史订单数据,构造超图G,使G中不同的顶点对应不同的商品,G中的超边则反映了商品在订单中的共现关系,通过对G做谱图分解,获得了对各商品的嵌入表示。然后,基于上述表示,提出了带上、下限簇数限制的约束聚类算法,并通过该算法生成商品分仓的结果。使用主题模型LDA(latent dirichlet allocation)生成的模拟订单数据作为实验分析的对象,在不同的参数设置下,生成了大量模拟订单数据,把本文提出的模型与当前流行的模型分别应用于这些数据以生成商品分仓的结果,实验结果显示,与现有的模型相比,本文提出的模型有明显的优势。

关键词: 商品分仓, 超图, 嵌入表示, 聚类, 运作优化

Abstract:

The rapid development of e-business(EB for short) requires EB companies to respond to consumer requirements quickly, in order to improve their efficiency of customer services, more and more EB companies have deployed distributeddistribution center systems. A main problem with using this system is that improper commodity warehouse allocation may result in a large number of orders being divided into multiple sub-orders, which not only causes inconvenience to customers, but also increases more operating costs to companies. Addressing to this problem, a commodity warehouse allocation model aimed at minimizing the total number of split orders is proposed. The model is based on a hypergraph representation of historic orders, where each vertex corresponds to a distinct goods appearing in the orders, and each hyperedge describes a co-occurrence relation of goods in a same orders. The spectrum decomposition on the Laplace matrix of the hypergraph is performed to obtain the embedding representations of the goods, then a constrained clustering algorithm is proposed and is applied to the above embeddings to generate the allocation results. The comprehensive simulations is conducted to evaluate the proposed model, where LDA (Latent Dirichlet Allocation) with variant parameter settings is used to simulate different ordering behaviors, and the proposed model as well as some other comparison models is applied to generate the allocation results from the data.All experimental results show the superior of our model.

Key words: commodity warehouse allocation, hypergraph, embedding representation, clustering, operations optimization

中图分类号: