地球信息科学学报 ›› 2018, Vol. 20 ›› Issue (11): 1541-1549.doi: 10.12082/dqxxkx.2018.180288

• 地球信息科学理论与方法 •    下一篇

基于自然语言空间关系描述的地图近似表达方法

曹青1,3,4(), 洪必文1,3, 张翎1,2,3,*(), 阮陵1,2,3, 龙毅1,3   

  1. 1. 南京师范大学地理科学学院,南京 210023
    2. 南京师范大学常州创新发展研究院,常州 213000
    3. 江苏省地理信息资源开发与利用协同创新中心,南京 210023
    4. 中国电信股份有限公司物联网分公司,南京 210023
  • 收稿日期:2018-06-19 修回日期:2018-10-03 出版日期:2018-11-20 发布日期:2018-11-20
  • 通讯作者: 张翎 E-mail:2454295083@qq.com;lingzhang.sky@gmail.com
  • 作者简介:作者简介:曹 青(1992-),女,硕士生,主要从事空间认知与地图可视化。E-mail: <email>2454295083@qq.com</email>
  • 基金资助:
    国家自然科学基金项目(41571382、61472191);江苏省高校自然科学研究重大项目(15KJA420001)

Map Approximate Expression Method Based on Spatial Relationship Description in Natural Language

CAO Qing1,3,4(), HONG Biwen1,3, ZHANG Ling1,2,3,*(), RUAN Ling1,2,3, LONG Yi1,3   

  1. 1. School of Geography Science Nanjing Normal University, Nanjing 210023, China
    2. Changzhou Institute of Innovation and Development, Nanjing Normal University, Changzhou 213000, China
    3. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
    4. China Telecom Company Limited, Internet of Things branch, Nanjing 210023, China
  • Received:2018-06-19 Revised:2018-10-03 Online:2018-11-20 Published:2018-11-20
  • Contact: ZHANG Ling E-mail:2454295083@qq.com;lingzhang.sky@gmail.com
  • Supported by:
    National Natural Science Foundation of China, No.41571382, 61472191;College Natural Science Research Key Program of Jiang Su Province, No.15KJA420001.

摘要:

自然语言和地图都具备表达地理实体空间关系的能力,自然语言使用方便、抽象化程度高,而地图更为直观,从自然语言转换到地图,有助于人们更深入地了解自然语言描述的地理实体空间关系。然而,如何让计算机具有从自然语言构建图形信息的能力,使计算机具有智能化空间认知思维是当前研究的难点。本文总结了自然语言空间关系描述的类型及特点,提出了基于自然语言描述的地理实体抽象表达方法以及空间关系近似转换方法,建立了一种基于自然语言空间关系描述的地图近似表达策略。实验结果表明,本文方法有效可行,能够实现定性描述的自然语言空间关系向定量(或近似定量)的图形空间关系的转换,为自然语言到地图的转换研究奠定了基础。

关键词: 自然语言, 地理实体, 定性空间关系, “文-图”转换;, 近似表达

Abstract:

With the further development of mobile GIS, intelligent GIS and socialized GIS, the geospatial information service based on natural language processing is an inevitable trend in the field of geographical information science. The intelligent conversion from text to map is one of the important research directions. Both natural language and maps have the ability to express spatial relationship of geographical entities. Natural language has the natural characteristic of usability and is highly abstract, while map language is more intuitive and revealing. The ubiquitous natural language contains a great deal of geographic information. Converting natural language to map language can help people intuitively understand the geographic space environment and bring out new discoveries. The current research difficulties focus on that how to make a computer construct graphical information from natural language and have the intelligent spatial cognitive thinking ability. This paper proposes a method that using point coordinated pairs, straight line segments and rectangular/circular shapes to quantitatively represent point, polyline and polygon geographical entities in natural language respectively. First the spatial relations description types in natural language between point and point, point and line, point and surface, line and line, line and surface, surface and surface geographic entities are summarized. Second, approximate transformation model of spatial relationships in natural language which considering the geometric types of geographical entities is constructed, and an approximate expression strategy based on spatial relationships description in natural language is proposed. Third, a prototype system is designed to implement "text-map" conversion, and scenic spot travel notes are selected as the experimental text to finish the experiment. The experimental results showed that the method mentioned above was feasible, the goal that converting qualitative spatial relationships in natural language to quantitative (or approximately quantitative) graphical spatial relationships could be achieved. This paper lays a foundation for the study of the conversion from natural language to map.

Key words: natural language, geographical entity, qualitative spatial relationship, "text - map" conversion, approximate expression