Chapter Preview
TopIntroduction
Geographic Information Systems (GIS) provide methods of capturing, storing, manipulating, analysing, and presenting different types of geographically referenced information. In conjunction with other areas of study, such as cartography and remote sensing, GIS have evolved into a discipline with its own research base known as Geographical Information Science (GIScience). Researchers in GIScience have increasingly recognised the need to consider time as an intrinsic element of models of geographic space, allowing the representation of dynamic elements in GIS. In GIScience, different approaches have been developed to dealing with elements which are subject to change over time, and an assorted terminology has been applied (e.g. geo-phenomena, dynamic GIS, spatio-temporal GIS).
Knowledge Representation (KR) is a sub-area of Artificial Intelligence (AI) which aims to represent elements and facts within a domain of knowledge through formal logic apparatus, in order to achieve intelligent behavior by providing methods of reasoning and drawing conclusions. The GIScience community has recently realised the role that Knowledge Representation (KR) approaches play in the development of modern GIS. Thus this article gives a discussion on KR approaches to modelling geographic phenomena, with particular attention to those based on the modeling of the concepts of events and processes.
This is a broad topic of research, with many interrelated areas, and is still under rapid development. Thus inevitably we shall not be able to cover all relevant work or even everything to the same level of detail. Rather, this article intends to give an overview on the main controversies, issues and challenges encountered in the field and present some of the approaches to overcome them. Relevant works in the area are left as additional readings. The next section presents background information on the topic. Then the following section gives an overview on existing approaches to representing and reasoning about events and processes and the concepts they are related to. Then the following sections, respectively, concludes the article and points to future direction in the field.
TopBackground
This section begins by introducing the field of KR, its related task of ontology development, and the problem of ontology grounding. Then an overview is given on qualitative spatial and temporal reasoning as important instruments for providing a KR of events and processes. Finally, this section gives an overall description of two phenomena which may affect the representation of events and processes: vagueness and information granularity.
Knowledge Representation
Knowledge Representation (KR) is one of the central areas in AI, and aims to provide appropriate representations of elements and facts within a domain of knowledge (Ringland & Deuce, 1988). These elements are represented as a set of symbols and logic is applied to provide formal apparatus to facilitate inferencing among these knowledge elements, resulting in new elements of knowledge. The fundamental goal of KR is to achieve intelligent behavior by providing methods of reasoning and drawing conclusions.
In the field of KR, substantial amounts of work focus on the development of ontologies. Ontology is defined by Gruber et al. (1993) as “an explicit specification of a conceptualisation” (p. 01). A conceptualisation, in turn, is defined as “an abstract, simplified view of the world that we wish to represent for some purpose” (Gruber et al., 1993, p. 01). It comprises “the objects, concepts, and other entities that are assumed to exist in some area of interest and the relationships that hold among them” (Gruber et al., 1993). Therefore, ontology deals with questions concerning what can be said to exist, and how such entities can related to each other. Ontologies can be specified in a wide variety of languages and notations (e.g., first-order logic, prolog, LISP). The ontology of geospatial phenomena is a critical emerging research theme in GIScience (Egenhofer, 2004). According to Kuhn (2005), geospatial ontologies can be seen as ‘GIS at the type level’. That is, they should provide reasoning capabilities (spatial and non-spatial) about conceptual geospatial elements, similar to the reasoning provided by GIS about their instances.
Key Terms in this Chapter
Ontology: Method of formally representing knowledge as a set of concepts within a domain, and the relationships which hold between them.
Geographic Information Systems (GIS): A type of system designed to capture, store, manipulate, analyse, manage, and display all forms of geographic information.
Qualitative Spatial Reasoning: Field of study in computer science whose goal is to provide ways of reasoning about space without the need for precise quantitative information.
Knowledge Representation: Area of artificial intelligence which aims to represent logically elements and facts within a domain of knowledge.
Vagueness: Phenomenon which arises from the lack of definite criteria to determine whether certain liguistic terms are appropriate to provide accurate descriptions of elements in the world.
Qualitative Temporal Reasoning: Field of study in computer science whose goal is to provide ways of reasoning about time without the need for precise quantitative information.
Ontology Grouding: Process of establishing an explicit link between ontology and data, by defining a set of primitive concepts/symbols of the ontology in terms of values and measurements, so that its concepts and relations become automatically recognisable from the data and their semantics are maintained concise throughout the changes occurring at data level.
Topology: Field of study in the area of mathematics concerned with basic properties of space, such as the connectedness.
Geographical Information Science (GIScience): Discipline with its own research basis which has emerged from the cooperation of GIS and other areas of study, such as cartography, remote sensing, photogrammetry, global positioning systems and geography.