A prototype auto-human support system for spatial analysis

【摘要】 <正>Spatial analysis is a multidisciplinary field that involves multiple influential factors, variation and uncertainty, and modeling of geospatial data is a complex procedure affected by spatial context, mechanism and assumptions. In order to make spatial modeling easier, some scholars have suggested a lot of knowledge from exploratory data analysis (EDA), specification of the model, fitness and diagnosis of the model, to interpretation of the model. Also an amount of software has improved some functionalities of spatial analysis, e. g. EDA by the dynamic link (GeoDa) and robust statistical calculation (R). However, there are few programs for spatial analysis that can automatically deal with unstructured declarative issues and uncertainty in machine modeling using the domain knowledge. Under this context, this paper suggests a prototype support system for spatial analysis that can automatically use experience and knowledge from the experts to deal with complexity and uncertainty in modeling. The knowledge base component, as the major contribution of the system, in support of the expert system shell, codes and stores declarative modeling knowledge, e.g. spatial context, mechanisms and prior knowledge to deal with declarative issues during the modeling procedure. With the open architecture, the system integrates functionalities of other components, e.g. GIS’ visualization, DBMS, and robust calculation in an interactive environment. An application case of spatial sampling, design and implementation of spatial modeling with such a system is demonstrated.更多还原

【Abstract】 Spatial analysis is a multidisciplinary field that involves multiple influential factors, variation and uncertainty, and modeling of geospatial data is a complex procedure affected by spatial context, mechanism and assumptions. In order to make spatial modeling easier, some scholars have suggested a lot of knowledge from exploratory data analysis (EDA), specification of the model, fitness and diagnosis of the model, to interpretation of the model. Also an amount of software has improved some functionalities of spatial analysis, e. g. EDA by the dynamic link (GeoDa) and robust statistical calculation (R). However, there are few programs for spatial analysis that can automatically deal with unstructured declarative issues and uncertainty in machine modeling using the domain knowledge. Under this context, this paper suggests a prototype support system for spatial analysis that can automatically use experience and knowledge from the experts to deal with complexity and uncertainty in modeling. The knowledge base component, as the major contribution of the system, in support of the expert system shell, codes and stores declarative modeling knowledge, e.g. spatial context, mechanisms and prior knowledge to deal with declarative issues during the modeling procedure. With the open architecture, the system integrates functionalities of other components, e.g. GIS’ visualization, DBMS, and robust calculation in an interactive environment. An application case of spatial sampling, design and implementation of spatial modeling with such a system is demonstrated.更多还原