Knowledge engineering is the process used in KEL to acquire and structure information about a subject. This approach serves to guide development of the integrated knowledge base required for problem-solving and decision making and identify deficiencies in the knowledge base needed for planning new research. Knowledge engineering methodologies have broad-based applications in both landscape ecological science and management
Problem-Solving and Decision Making:
Contemporary issues in landscape ecological science and management all have a substantial existing knowledge base associated with them. Problem-solving and decision making begin with effective and efficient use of this knowledge. KEL specializes in development of computer applications to organize, integrate, and interpret existing knowledge. The goal is to use t he full measure of information available for problem-solving and decision making. Typically, domain-specific information from a variety of academic disciplines (ecology, geography, forestry, psychology, and sociology) forms the knowledge base for an environmental problem. Tools and techniques from engineering and computer science are needed to integrate the various elements of the knowledge base, so that it can be used in an efficient and effective manner. The various kinds of computer-based systems available to support landscape problem-solving and decision making are illustrated in Figure 1. Emphasis in KEL has been placed on object-oriented simulation, expert systems, knowledge-based systems, intelligent geographic information systems, and the knowledge system environment. These techniques permit use of spatially referenced data a, tabular information, and heuristic knowledge of technical experts. They are useful for scientific study of landscapes and applied problems in land-use management.
Figure 1. Computer decision aids for problem-solving and decision making in environmental science and management.
Knowledge Engineering In Landscape Management
Landscape management deals with subjects that typically have large and disparate knowledge bases. The data and information that form the knowledge base for a specific problem often come from several different domain specialties, e.g., ecology, geography, sociology, economics. The knowledge base can exist in several forms: (i) tabular information [usually stored in a database management system], (ii) spatially referenced data themes [usually associated with a geographic information system - GIS], (iii) numerical output from simulation models and mathematical evaluation functions, and (iv) heuristics of experts [based on corporate experiences of humans]. Although the knowledge base for most problems in landscape management is substantial, it is also incomplete and in a state of evolution. Knowledge engineering is an activity that embraces a set of concepts and methodologies dealing with (i) acquisition of knowledge, (ii) analysis and synthesis of data and information [quantities], (iii) integration and interpretation of knowledge [quantities and qualities], and (iv) application of knowledge (Figure 2). The goal of this activity, in the context of landscape management, is to facilitate use of the full extent of knowledge available for the purpose of solving a problem, supporting decisionmaking, or developing a plan of action. Historically, scientists have also used personal flavors of this basic approach to conduct, summarize, and report their research discoveries. Computer-based tools and technologies have been created to formalize and automate the process, thus greatly expanding human capabilities. Knowledge engineering can be viewed as a computational approach to landscape management. Each of the elements of knowledge engineering is briefly described below. The elements include acquisition of knowledge, analysis and synthesis of data and information, and integration and interpretation of knowledge.
Acquisition of Knowledge
Every program in landscape management requires an evaluation of the extant data and information that form the knowledge base for a specific problem. There are three basic activities associated with the knowledge acquisition process: definition, elicitation, and appraisal. Definition of relevant data and information follows from a systematic evaluation of the problem of interest. Facilitation tools (e.g., Object Oriented Program Planning(tm)) can guide the formulation of specific project objectives. Solution pathways can then be created. Elicitation deals with acquiring information directly from experts. The goal is to develop a formal knowledge base for a particular topic or problem. A variety of techniques has been devised to guide the process, e.g., focused interview, structured interview, probes, goal decomposition, etc. Computer-based tools are available to assist in knowledge base construction, maintenance, and documentation (e.g., Netweaver(tm)). Appraisal deals with evaluation of the data and information that form the knowledge base for a particular problem. All scientists perform this task when they seek to place the results of their research into the corpus of existing knowledge. For complicated problems, where a variety of sources and types of data and information are involved (e.g., evaluating the impact of global warming on biodiversity), computer-based systems (e.g.,Netweaver(tm)) are extremely useful for ordering and organizing extant data and information. These systems are also useful in identifying deficiencies in knowledge.
Analysis and Synthesis of Data and Information
Ecology is a science where the objects of interest or study (plants, animals, the elements of the environment) can be described by units and dimensions, e.g., biomass in g/m2. Interpretation of results from ecological research usually involves the analysis or synthesis of data and information represented as scaled quantities, i.e., objects defined by units and dimensions. Although the subjects of analysis and synthesis are often discussed in the same context, they are fundamentally different activities. Furthermore, the tools and techniques used are different. Analysis deals with separating or breaking up of any whole into its parts so as to find out their nature, proportion, function, or relationship, e.g., analysis of variance. There are three common approaches to analysis: graphical, numerical, and statistical. Each of these approaches can be further sub-divided. For example, elements of statistical analysis include: environmental design and sampling, spatial statistics, statistical ecology, environmental regulatory statistics, environmental monitoring, and environmetrics. Analysis of categorical map information is a central focus of KEL and a variety of approaches are utilized (Fig. 3.) Synthesis deals with the putting together of parts or elements so as to form a whole, i.e., it is the antithesis of analysis. There are three common approaches: simulation, optimization, and visualization. Each approach represents a substantial and well-developed discipline that can be further compartmentalized. For example, simulation can be viewed as continuous (systems of differential equations) or discrete (object-oriented simulation). Optimization (in relation to restrictions) includes linear programming, non-linear programming, dynamic programming, and control theory. Visualization includes graphics, animation, and four-dimensional representations.
Integration and Interpretation
Although scientists prefer to deal with quantities, managers rely also on qualitative judgments based on their experience (or the experiences of others), i.e., they use heuristic knowledge as well as quantitative data and information to solve problems, make decisions, and develop plans. Furthermore, scientific understanding of a specific landscape management problem will rarely be so complete as to eliminate the need for qualitative assessment and human judgment. Therefore, our emphasis at KEL centers on integrative systems that are useful for blending quantitative as well as quantitative information. The integrative systems are based on technologies adapted from a variety of subject domains, e.g., computer science, engineering, mathematics, cognitive psychology, management science, etc. Several types of integrative systems have been used to address ecosystem management problems (Figure 1): expert systems, intelligent geographic information systems, intelligent database management systems, object-oriented simulation, and knowledge-based systems. Note that the increasing complexity of the computer-based tools parallels the levels of human comprehension. The advanced levels (experience processing, shared visions, and epiphanies) (Figure 1) can only be achieved through the use of the full measure of knowledge available on a subject. The computer-based system that addresses this challenge is the knowledge system environment (KSE).