计算机与现代化 ›› 2022, Vol. 0 ›› Issue (06): 43-48.

• 数据库与数据挖掘 • 上一篇    下一篇

C2C在线短租跨平台房东匹配算法

  

  1. (广东工业大学管理学院,广东广州510520)
  • 出版日期:2022-06-23 发布日期:2022-06-23
  • 作者简介:吴代漾(2000—),男,广东汕尾人,本科生,研究方向:数据挖掘,E-mail: 1821203795@qq.com; 通信作者:赵洁(1979—),女,广东肇庆人,教授,博士,研究方向:机器学习,数据挖掘算法设计,智能匹配,商务智能,E-mail: zhaojie@gdut.edu.cn; 梁家铭(1977—),男,广东广州人,硕士研究生,研究方向:数据挖掘算法设计,机器学习,E-mail: 2111908031@mail2.gdut.edu.cn; 董振宁(1977—),男,河南洛阳人,副教授,博士,研究方向:供应链管理,供应链金融,物流园区规划,物流信息系统,E-mail: 5393966@qq.com; 梁周扬(1985—),男,广东阳江人,实验师,博士,研究方向:移动商务,推荐系统,E-mail: 342791386@qq.com。
  • 基金资助:
    国家自然科学基金资助项目(71871069); 教育部人文社会科学研究规划项目(18YJAZH137).

Host Matching for C2C Online Short-term Rentals

  1. (School of Management,Guangdong University of Technology, Guangzhou 510520, China)
  • Online:2022-06-23 Published:2022-06-23

摘要: 随着民宿与在线短租平台的兴起,房东多归属现象持续受到关注与研究,该现象提供了新的研究角度,而如何在不同平台识别同源房东成为首要解决的问题。故本文基于传统用户匹配探索C2C在线短租跨平台房东匹配算法。其中由于房东个人信息稀疏,因此本文引入房源信息,设计基于房源信息的两阶段房东匹配算法(TSHM)。本文方法在基于国内2大在线短租平台真实数据划分的普通数据集与难例数据集上分别达到99.69%与81.97%的准确率,优于SVM、DT等传统分类器,验证了匹配模型与匹配特征的有效性,为跨平台房东匹配提供新思路,在房东个人信息缺乏条件下仍可有效匹配房东。但本文仅针对国内平台数据进行实验,未引入文本与图片等特征,存在一定局限性。

关键词: 在线短租, 房东匹配, 遗传算法, 多归属

Abstract: With the rise of homestays and online short-term rental platforms, the phenomenon of host multiple ownership continues to receive attention and research. This phenomenon provides a new research perspective, and how to identify same-source hosts on different platforms has become the first problem to be solved. Therefore, this article explores the C2C online short-term rental cross-platform host matching algorithm based on traditional user matching. Among them, due to the sparse personal information of the host, this paper introduces housing information and designs a two-stage host matching algorithm (TSHM) based on housing. The method in this paper achieves 99.69% and 81.97% accuracy on the common data set and the hard-case data set based on the real data of the two domestic online short-term rental platforms, respectively, which is better than traditional classifiers such as SVM and DT. The matching model is verified. The effectiveness of the matching features provides a new idea for cross-platform host matching, which can still effectively match the host even if the host’s personal information is lacking. However, this article only conducts experiments on domestic platform data, and does not introduce features such as text and pictures, which has certain limitations.

Key words: online short-term rent, host matching, genetic algorithm, multi-homing