Intelligent spatial decision support systems (Q1381078)

This book covers a broad field of conceptual, theoretical, and system development perspectives of intelligent spatial decision support systems. The author starts from declarative and procedural knowledge definitions and pays great attention to basic knowledge representation formalism and associated inference. He outlines the foundation of the propositional and predicate logic and to the main rules of inference as modus ponens and the rule of universal instantiation. He introduces production system, semantic network, frame and object oriented approach for knowledge representation and inference. The whole chapter is devoted there to fuzzy logic approaches, what comprises modelling of uncertainty using fuzzy subsets and fuzzy relation. Great attention is paid to inference rules of fuzzy logic. If the fuzzy proposition is examined very carefully from the point of fuzzy inference and the same attention is paid to quantified proposition and associated inference rules. The author continues with uncertainty management and considers two types of uncertainty, namely uncertainty due to randomness and uncertainty due to imprecision. Several systems of inference rules with uncertainty measures are examined. He outlines the measure of belief and disbelieve approach with certainty factor, the belief and plausibility approach and possibility and necessity approach. The next chapter concerns neural network approaches to spatial knowledge representation and inference. An introduction to neural network theory with both feedforward and feedback networks is given. It is shown, how to construct one layer feedforward networks, which model simple logical operations. Within this chapter, he explains the learning process in feedforward neural networks. A part of the chapter is focussed on special types of networks, which are examined from the point of autoassociative and heteroautoassociative memories. The next topic, which is studied, is genetic algorithm. The author explains the frame and the basic operations of genetic algorithm and gives a concise overview of the theory concerning canonical genetic algorithm. He mentions two systems, which employ genetic algorithms. The first of them is SCION, which enables automatically generate, test and select good rules for classification. The second system is called GENNET and has been designed to optimize the structure of a neural network. In the remainder of this book, the author deals with quadratic line how to store and process data concerning space. He describes vector and raster data models and the possibility to record topological properties of individual objects. He introduces databases as the main structure of data storage and mentiones various types of database languages. Further, the author examines various ways of network representation from the perspectives on how the associated data should be exploited and discusses reasons for an expert system to be embedded into a development tool for spatial decision support system. One application into land classification problem is mentioned there and attention is paid to the graphical display of imprecision. As one of the possible applications, an estimation of damages caused by one 50-years rain storm in watershed of one Chinese province is given. The conclusion of the book is devoted to the object-oriented expert system shell for building spatial decision support systems in connection with tools for multiobjective compromising.