Skip to main content
main-content

Über dieses Buch

In the past half century, we have experienced two major waves of methodological development in the study of human behavior in space and time. The fIrst wave was the well known "quantitative revolution" which propelled geography from a mainly descriptive discipline to a scientifIc discipline using formalism such as probability, statistics, and a large-number of mathematical methods for analyzing spatial structures and processes under certainty and uncertainty. The second wave is the recent advancement of geographical information systems which equips geographers with automation in the storage, retrieval, analysis, and display of data. Both developments have significant impacts on geographical studies in general and solutions to real life spatio-temporal problems in particular. They have found applications in urban and regional planning, automated mapping and facilities management, transportation planning and management, as well as environmental planning and management, to name but a few examples. Both developments have one thing in common. They one way or the other use computer to process and analyze data. However, not until recently, there has been very little interaction between the two. Quantitative models have largely been developed independent of the underlying data models and structures representing the spatial phenomena or processes under study. Display of analysis results has been primitive in terms of the utilization of computer graphic technologies. Formal models, in addition to their technical difficulties, have poor capability in communication with users. Geographical information systems, on the other hand, have originally been developed with a slight intention to entertain powerful analytical models.

Inhaltsverzeichnis

Frontmatter

1. Introduction

Abstract
Decision making in space and time is a highly complex process of choosing among alternatives to attain an objective or a set of objectives under constraints. It can be a structured process involving problems with standard solution procedures, or an unstructured process consisting of problems with no clear-cut solution procedures, or even semi-structured problems for which combinations of standard procedures and individual judgements have to be used to find a solution. All these processes generally involve voluminous spatial and aspatial information, structured and unstructured knowledge, human valuation and judgement.
Yee Leung

2. Symbolic Approaches to Spatial Knowledge Representation and Inference

Abstract
Knowledge representation and inference are main concerns in building systems with artificial intelligence. To be able to understand and to reason, an intelligent machine needs prior knowledge about the problem domain. To understand sentences, for example, natural language understanding systems have to be equipped with prior knowledge about topics of conversation and participants. To be able to see and interpret scenes, vision systems need to have in store prior information of objects to be seen. Therefore, any intelligent systems should possess a knowledge base containing facts and concepts related to a problem domain and their relationships. There should also be an inference mechanism which can process symbols in the knowledge base and derive implicit knowledge from explicitly expressed knowledge.
Yee Leung

3. Fuzzy Logic Approaches to Spatial Knowledge Representation and Inference

Abstract
In Chapter 2, knowledge is assumed to be exact and a statement or an inference is either true or false. However, human knowledge is often inexact and our inference often consists of a certain level of uncertainty. While uncertainty is of various sources (Graham and Jones, 1988; Klir, 1988; Leung, 1988a), the one stems from imprecision is rampant in human systems. To represent and infer with such knowledge, we need a logical system which can handle imprecision. Among existing paradigms, fuzzy logic appears to be instrumental in processing imprecision in SDSS.
Yee Leung

4. Management of Uncertainty in Knowledge Representation and Inference

Abstract
In any inference, a certain amount of uncertainty is involved. In this Chapter, I do not intend to discuss the fine details of the philosophical arguments about uncertainty (for some detailed breakdowns of uncertainty, see for example Graham and Jones, 1988; Klir and Folger, 1988; Smithson, 1989). I, however, concentrate on the management of two major types of uncertainty in knowledge representation and inference. They are uncertainty due to randomness and uncertainty due to imprecision (see Leung, 1988a, Chapter 2, for a discussion).
Yee Leung

5. Neural Network Approaches to Spatial Knowledge Representation and Inference

Abstract
Our discussion so far has concentrated on the symbolic approaches to spatial knowledge representation and inference. Logic (fuzzy and non-fuzzy), production systems, semantic networks, frames, object-oriented programming, and their hybrids all belong to symbolic systems in which knowledge is modeled by symbols. Intelligence is realized by a symbolic structure in which symbols can be manipulated and reasoning can be made. The advantages of the symbolic approaches are that they provide a structured representation of knowledge so that processing elements corresponding to meaningful concepts and inference can be traced and explained. The separation of knowledge from the inference mechanism also makes knowledge update easier and more efficient. The approach is thus a top down approach which gives consensus knowledge to a system by instructing it what to feel and respond without having to gain knowledge through experience. It may be a faster way to build intelligent system.
Yee Leung

6. Knowledge Acquisition for Spatial Inference — The Case of Genetic Algorithms

Abstract
Intelligent SDSS integrating human expertise and very large spatial databases have become a necessity for solving our highly complex spatial problems such as resource exploration, land-type classification, transportation planning, and environmental management. In addition to a sound theoretical foundation, a good system design, and a powerful and user-friendly software environment, the success of intelligent SDSS lies heavily on the domain specific knowledge they acquired. Knowledge acquisition thus plays a very important role in the design of intelligent SDSS in general and spatial expert systems in particular (Barbanente et al., 1992).
Yee Leung

7. Spatial Data Models and Data Structures

Abstract
Similar to knowledge representation, data handling is an important function in SDSS. It involves the ways spatial and aspatial information is conceptualized, structured, and implemented in computers. Literature on data models and data structures is voluminous (see for example Brackett, 1987; Decker, 1989; Modell, 1992). Geometric data models and data structures have also been studied in GIS research over the years (see for example Peuquet, 1984; Burrough, 1986; Vaughaneia et al., 1988; Maguire et al., 1991; Laurini and Thompson, 1992). The purpose of this chapter is not to provide an overview of research in data models and data structures in general and GIS in particular. I concentrate instead on the modeling and structuring of geometric and attribute data, especially under uncertainty, in SDSS.
Yee Leung

8. Management of Models in Spatial Decision Support Systems

Abstract
To solve a large variety of spatial problems, researchers have developed over the years numerous structured models, in the form of statistical methods, mathematical models, heuristic procedures, and algorithms, for decision making involving human and physical processes. Differing from declarative knowledge discussed in Chapter 2, this type of knowledge has a highly structured format with rigid execution procedures. They are instrumental in formulating and solving problems with well specified conditions, structures, and mechanics. Their formal logic and solution methods unfortunately are usually difficult or too time consuming for decision makers to understand. Moreover, these models come in such a variety so that non-technical decision makers generally get confused or unsure in determining under what conditions they could appropriately be applied. Inappropriate or insufficient utilization of this type of knowledge has thus hampered an effective and efficient decision making process.
Yee Leung

9. An Expert System Shell for Building Spatial-Decision-Support-System Development Tool

Abstract
I discuss in this chapter a spatial-decision-support-system development tool (shell) which we have developed, implemented and applied to construct SDSS to assist spatial decision-making tasks (Leung et al., 1995a, b). The shell is a general SDSS development environment constructed for the purpose of building SDSS to solve specific spatial problems in an effective and efficient manner. The overall objective is to have a system which can materialize the general SDSS architecture depicted in Fig. 1.2 in Chapter 1.
Yee Leung

10. A Spatial Decision Support System for Flood Simulation and Damage Assessment Using FLESS

Abstract
In order to illustrate how FLESS can be employed to build domain specific SDSS, I describe in this section a flood simulation and damage assessment problem which requires the integrative utilization of mathematical models (structured knowledge), rules (unstructured knowledge), and GIS in interactive decision support. The study area is the watershed of Sun Hugou (a river network) located in the central mountain area of Huairou County, north of Beijing, China.
Yee Leung

11. An Object-Oriented Expert System Shell for Building Spatial Decision Support Systems

Abstract
Similar to building object-oriented DBMS and MBMS, expert system shell can be developed within an object-oriented framework. Such a shell is instrumental in constructing tools for building object-oriented SDSS. Parallel to the function of FLESS, an object-oriented expert system shell can serve as the nerve center of a SDSS. A full-fledged object-oriented SDSS tool will have the DBMS and MBMS developed within an object-oriented environment.
Yee Leung

12. Conclusion

Abstract
I have discussed from the conceptual, theoretical, and systems development perspectives various issues of intelligent SDSS in this monograph. Examples and applications have also been extensively employed to substantiate the conceptual arguments. It is apparent that an intelligent SDSS should possess the capability of handling interactively and effectively knowledge (structured and unstructured), information (spatial and aspatial), and human–machine interaction. It should be able to assist decision making by providing alternatives through the efficient processing of knowledge and information in a complex environment. The system should also be robust and flexible in managing uncertainty of various types. Furthermore, it should be responsive to users’ demands.
Yee Leung

Backmatter

Weitere Informationen