地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (1): 16-28.doi: 10.12082/dqxxkx.2021.200632

• 地球信息科学综述 • 上一篇    下一篇

面向遥感大数据的地学知识图谱构想

王志华1,2(), 杨晓梅1,2,*(), 周成虎1,2   

  1. 1.中国科学院地理科学与资源研究所 资源与环境信息系统国家重点实验室,北京100101
    2.中国科学院大学,北京100049
  • 收稿日期:2020-10-22 修回日期:2020-12-09 出版日期:2021-01-25 发布日期:2021-03-25
  • 通讯作者: 杨晓梅
  • 作者简介:王志华(1988— ),男,河南信阳人,副研究员,主要从遥感地学分析研究。E-mail: zhwang@lreis.ac.cn
  • 基金资助:
    国家自然科学基金项目(41901354);国家自然科学基金项目(41671436);国家自然科学基金项目(41890854)

Geographic Knowledge Graph for Remote Sensing Big Data

WANG Zhihua1,2(), YANG Xiaomei1,2,*(), ZHOU Chenghu1,2   

  1. 1. State Key Laboratory of Resources and Environment Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2020-10-22 Revised:2020-12-09 Online:2021-01-25 Published:2021-03-25
  • Contact: YANG Xiaomei
  • Supported by:
    National Natural Science Foundation of China(41901354);National Natural Science Foundation of China(41671436);National Natural Science Foundation of China(41890854)

摘要:

由于地球表面的时空异质性与复杂性,传统从遥感影像具有的信息特征出发,构建智能解译算法解决遥感地学认知的思路在应对面向全球的海量遥感大数据分析时,其精度和地学实用性已触及瓶颈。为此,本文从地学知识为核心的角度出发,结合当前知识图谱理论的发展,提出一种新的面向遥感大数据分析的地学思维构想——地学知识图谱。本构想将地学知识的概念进行重构,依次划分为数据性知识、概念性知识和规律性知识3个层次,并分别利用图模型的节点和边进行统一化表达和关联,打通不同层次地学知识间的反馈迭代与更新,在此基础上赋予地学知识图谱分析遥感大数据分析时知识的查询检索、知识推理、动态校正、拓展更新等功能。其中,如何构建具有多尺度、高维度特征的地理实体以及大体量、异质性的知识层级间的关联推理是地学知识图谱构想实现的关键难点。得益于知识的分层次和图模型结构的统一化表达,提出的地学知识图谱构想在促进遥感大数据时代背景下的地学知识精准化,提升遥感大数据解译精度和地学实用性,深化地学规律认知等方面应该具有广阔的前景。

关键词: 遥感大数据, 遥感信息提取, 遥感智能解译, 土地利用/覆盖变化, 地学知识图谱, 地学信息图谱, 地学知识, 知识图谱

Abstract:

Due to the temporal and spatial heterogeneity of the complex earth's surface, the traditional idea of developing new intelligent interpretation algorithms to solve the remote sensing geoscience cognition based on the features of remote sensing images has hit the bottleneck in terms of accuracy and geographic usage when analyzing remote sensing big data. To overcome the bottleneck, we proposed the Geographic Knowledge Graph (GKG) that based on the geographic knowledge to analyze the remote sensing big data, which is inspired by the recently proposed Knowledge Graph from the geographic perspective. It expands the concept of the geographic knowledge and classifies the geographic knowledge into three levels: Data knowledge, conception knowledge, and regularity knowledge. Then, it represents and connects all geographic knowledge in Graph by nodes and edges and realizes the feedback iteration and update between different levels of the geographic knowledge. This representation enables GKG to perform well at knowledge inquiring, reasoning, calibration, and expanding. How to construct multiscale high-dimension geo-entities and how to connect different levels of the geographic knowledge with heterogeneous features are two key technologies. These functions make GKG promising in refining existing geographic knowledge in the era of remote sensing big data, promoting remote sensing interpretation accuracy and geographic usage, and promoting the development of geoscience.

Key words: remote sensing big data, remote sensing information extraction, remote sensing intelligent interpretation, land use/cover change, geographic knowledge graph, geo-information Tupu, geographic knowledge, knowledge graph