地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (1): 93-103.doi: 10.12082/dqxxkx.2021.200424

• 专栏:"全空间信息建模分析方法与应用研究" • 上一篇    下一篇

网络地理信息服务中用户空间访问聚集行为研究

陈文静1(), 李锐1,*(), 董广胜1, 李江2   

  1. 1.武汉大学测绘遥感信息工程国家重点实验室,武汉 430079
    2.湖北省自然资源厅信息中心,武汉 430071
  • 收稿日期:2020-07-31 修回日期:2020-09-30 出版日期:2021-01-25 发布日期:2021-03-25
  • 通讯作者: 李锐
  • 作者简介:陈文静(1996— )女,安徽宣城人,硕士生,主要从事网络地理信息服务用户行为研究。E-mail: chenwenjing@whu.edu.cn
  • 基金资助:
    国家重点研发计划项目(2016YFB0502301);国家自然科学基金项目(41771426)

Research on User Spatial Access Aggregation Behavior in Network Geographic Information Service

CHEN Wenjing1(), LI Rui1,*(), DONG Guangsheng1, LI Jiang2   

  1. 1. State Key Laboratory of Information Engneering in Surveying, Mapping and Remote sensing, Wuhan University, Wuhan 430079, China
    2. Information Center of Department of Natural Resources of Hubei Province, Wuhan 430071, China
  • Received:2020-07-31 Revised:2020-09-30 Online:2021-01-25 Published:2021-03-25
  • Contact: LI Rui
  • Supported by:
    National Key Research and Development Foundation of China(2016YFB0502301);National Natural Science Foundation of China(41771426)

摘要:

研究网络地理信息服务用户的访问行为,有利于了解用户地理信息兴趣、实现按需服务。本文基于全空间信息系统建模的理论,构建用户-访问城市关系网络,研究用户访问的空间聚集性。顾及到关系网络中行为关系强度的表达需要同时考虑用户访问行为、城市关联关系和城市结构,仅用单一的用户访问行为数据会存在偏差,本文提出了基于矩阵分解的数据融合方法,对网络地理信息服务中用户访问数据、城市关联数据以及城市的POI(兴趣点)数据进行融合,表达用户-城市访问关联强度。在此基础上,基于关系网络聚类方法实现用户的聚集模式挖掘。考虑到只以空间距离实现聚类的方法无法兼顾关系网络中用户对不同城市的访问偏好特征,本文在FCM(模糊C均值聚类算法)的基础上以用户对城市的访问概率定义访问偏好提出PFCM算法,同时兼顾关系网络中城市间的空间距离和访问行为关系强度,减小聚类结果的偏差。本研究通过用户访问的空间聚类表达用户访问的空间兴趣偏好,有助于理解用户访问行为与城市之间的相互关系,为网络地理信息服务在数据缓存和提前推送等方面的性能提升提供指引,从而更好的服务于用户访问。

关键词: 全空间信息系统, 城市群, 用户访问行为, 访问偏好, 关系网络, 数据融合, 矩阵分解, 模糊聚类

Abstract:

The rapid development of the geographic information industry has promoted the popularization of network geographic information services, providing the public with indispensable and convenient services such as spatial positioning, spatial query, and path planning, which penetrate all aspects of life. At the same time, the number of users is also exploding, how to provide users with on-demand and high-quality geographic information services has become one of the key issues to be solved. So it is meaningful to study the access behavior of users of network geographic information services, which is conducive to understanding users' geographic information interests and realizing on-demand services. Based on the theory of full spatial information system modeling, this paper constructs a user-visited city relationship network and studies the spatial aggregation of user access. Users' behavioral relationship strength generally involves many factors. This paper takes into account that the expression of behavioral relationship strength in relational network needs to consider user access behavior, city association relationship, and city structure at the same time, and there will be bias on a single user accesses behavior data. In order to solve this bias, this paper proposes a data fusion method based on matrix decomposition to integrate user access data, city associated data, and Point of Interest (POI) data in network geographic information services to express the strength of user-city access correlation. In the relational network, the fusion data are used to express the strength of behavioral relationship, and the distance between cities is used to express the strength of spatial relationship. On this basis, the clustering pattern mining of users is realized based on the relational network clustering method. The characteristics of users' preference to different cities will affect the clustering results. Given that clustering methods are usually based on spatial distance to achieve clustering, they cannot take into account the user's preference characteristics of different cities in the relationship network. On the basis of Fuzzy C-means Clustering algorithm (FCM), this paper proposes the PFCM algorithm based on the user's access probability to the city definition of access preference. At the same time, the spatial distance between cities and the strength of access behavior relationship in the relational network are taken into account to reduce the deviation of clustering results. This research expresses the spatial interest preferences of users through the spatial clustering of user visits. It helps to understand the relationship between user access behavior and cities and provides guidance for the performance improvement of network geographic information services in data caching and advance push so as to better serve user access.

Key words: full-space information system, urban agglomeration, user behavior, access preferences, relationalnetwork, data fusion, matrix decomposition, fuzzy clustering