摘要
自然语言空间关系理解、计算与表达是空间分析、地图综合、人工智能等的基础和难点问题之一。between经常用于描述空间对象的空间排列,指其中一个中间对象位于其他两个空间对象之间,是其他自然语言空间谓词度量的基础。针对现有方法度量因子单一的不足,本文提出了改进的between关系的定量度量模型。首先,通过问卷调查发现影响人类认知between关系的因素——面积比、偏移量和空间对准程度;然后,设计了度量指标的计算方法,并提出了偏移量这一新指标的计算方法;最后,构建了between关系的定量度量模型,利用层次分析法自动确定3个因素的权重系数,实现between关系的定量度量。实验表明,该方法能够有效度量between关系,符合人类空间认知习惯。
Understanding,computing and expressing the spatial relationships in natural language constitutes a fundamental but challenging problem in spatial analysis,map generalization,artificial intelligence and also other related domains.The term“between”is frequently employed to denote the spatial arrangement of spatial objects,signifying that one of the intermediate objects is situated between the other two spatial objects.This spatial concept always serves as the foundation of the metric for other natural language spatial predicates.To address the shortcomings of existing methods in measuring a single factor,this paper proposes an improved quantitative metric model for the ternary spatial relationship“between”.First,a questionnaire survey was conducted to identify the factors affecting human cognition of the between relationship,which turned out to be area ratio,offset,and spatial alignment.Then,the calculation method of the three indicators was designed,and the offset index was innovatively proposed.Finally,the quantitative metric model of the relationship was constructed,and the weight coefficients of the three indicators were automatically determined using the Analytic Hierarchy Process.Experimental results demonstrate the effectiveness of the proposed method in measuring the“between”relationship,which aligns with human spatial cognition.
作者
张寒雪
巩现勇
钱海忠
刘呈熠
李振兴
武芳
ZHANG Hanxue;GONG Xianyong;QIAN Haizhong;LIU Chengyi;LI Zhenxing;WU Fang(Information Engineering University,Zhengzhou 450001,China)
出处
《测绘科学技术学报》
2025年第5期472-478,共7页
Journal of Geomatics Science and Technology
基金
国家自然科学基金面上项目(42371461)。
关键词
三元空间关系
定量度量
面积比
偏移量
空间对准
ternary spatial relationship
quantitative metrics
area ratio
offset
spatial alignment
