地理空间中的地物,根据其本体类型具有不同的地理语义特征。受到地理语义的影响,人们在描述几何特征和关系相似的物体之间的空间关系时,所用的词汇是不同的。目前,人们在空间关系的自然语言查询方面的研究主要集中在空间关系描述、空间关系自然语言形式化表达、空间关系自然语言查询语句的转换,以及空间关系自然语言查询的查询接口等方面,没有考虑到地理本体的语义差异,这会使得空间关系的自然语言查询不能符合人类语言和认知的习惯。本文依据一些学者在空间关系的自然语言描述的形式化(主要是线和线、线和面等形状的空间地物之间的关系)研究的基础上,针对不同地理本体类型的空间地物和空间关系自然语言描述词汇之间对应的语义关系有所不同的特点,总结了自然语言词汇在描述空间关系时的对应规则,提出不同特征类型地物和描述词汇之间的规则库设计方法,并设计部分空间地物的规则实例。最后,结合空间关系判断函数,设计具体查询系统实践举例证明。
The description of a spatial relation is the reflection of human' s cognition of spatial objects. It is not only affected by the topology and metric, but also affected by geographic semantics, such as the categories of geographic entities and contexts. Currently, the researches about language aspects of spatial relations mostly focus on natural-language formalization, parsing of query sentences, and natural-language query interface. In these researches, spatial objects are usually abstracted into geometric figures. However, geographic objects are not simple geometric points, lines or polygons. Spatial relation queries that just consider geometry property of objects cannot always meet human' s habits of language and cognition. So, in order to get a sound answer according with human cognition in spatial relation query, we have to take the semantic aspect of geographic objects into account. Based on previous work on natural-language formalization and human-subject tests, the functions of the natural-language terms which describe spatial relations are calculated in this paper. Then, we sum up some knowledge about how spatial relations between two different geographic objects with different geographic semantics are represented in nat- ural language, and design a knowledge base to represent these knowledge rules. Finally, using the rules in knowledge base and GIS platform, the prototype of a query system is implemented. The results show that spatial relation queries which take geographic semantics into account can be more in accordance with people's cognitive habits of geographic knowledge.