Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!

Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


1.3. TECHNOLOGICAL DEVELOPMENT

The narrow scope of the early systems did not call upon very sophisticated techniques. Most expert system developments followed the overall pattern using rapid prototyping and expert system shells. During the first years, expert systems became strongly associated with backward-chaining, rule-based programs founded on the MYCIN tradition (Shortliffe 76). This conception tends to linger in spite of new and powerful techniques that many of today's efforts capitalize on. This was also the situation for the system developments of the type discussed here. The development pattern reflected by the history of geological systems is typical for the whole field of expert systems.

Over the years, knowledge representation has developed and many new representational schemes have been explored. Standard inferences have been augmented by several new control paradigms, many of which use knowledge about the problem-solving process itself to govern the execution of the system. The use of a single knowledge base was the most common when interest in expert systems took off. At the end of the 1980s, systems using multiple knowledge bases were explored. Multi-agent systems and cooperating expert systems were frequently reported. Model-based versus purely heuristically based systems proved useful. Systems using a hybrid set of techniques were common at the start of the 1990s.

The first expert systems for management of natural resources were simple single-user, stand-alone programs. The focus of the development effort was the knowledge base, the inference engine, and the knowledge acquisition effort. User-friendliness was not a significant issue and integration into a wider computing environment was not addressed. Systems seeking operational status soon found themselves placed in a network or in a setting where they exchanged data with both databases, ordinary number crunching programs, simulators, and geographical information systems (GIS).

For many, the embodiment of knowledge in an expert system must be explicit. However, the expertise that can be implicitly captured in a neural network might well outperform human experts for both recognition tasks and interpretation. Developers within the oil domain became aware early on of the benefits posed by soft computing techniques related to neural networks, fuzzy systems, and genetic algorithms. Hence, they adopted these problem-solving techniques for purposes that traditional rule-based systems had failed or proven costly to develop. Today, many systems incorporate expertise dedicated for tasks that are especially suited for soft computing. This trend has manifested itself over the past 5 years. Soft computing techniques must therefore be considered part of the toolbox of the expert system developer.

Object-oriented systems development has had a significant effect on expert systems trends. Many of the systems presented here apply variants of object-oriented software techniques. The object-oriented paradigm promises very good models of the real world. It lends itself very well to explicit knowledge modeling and incorporates representational power shared by both semantic networks and frames. Other cherished features include the reusability of code and thus knowledge. In addition, the flexibility of message passing has made it easy to build separate knowledge agents capable of self-organizing and cooperative problem-solving.

Of particular interest are those new systems that apply old ways in new settings. The advent of the Internet and the WorldWideWeb has opened a new world for expert systems. Some development efforts demonstrate both existing capabilities as well as future potentials that will be highlighted here.

1.4. SCOPE

Our discourse covers a broad range of systems. But it is admittedly colored by the dominance of geology-related systems in the 1980s and early 1990s. This is due both to their abundance at the start and their significance in terms of technology development. Expert systems builders seeking new solutions for persistent problems related to exploration, oil well logging, and seismic interpretation worked shoulder-to-shoulder with computer scientists concerned with the fundamental aspects of artificial intelligence. Real-world tasks replaced the toy problems nursed and cherished in the university labs. This shift and the funds invested to tackle the new problems became a driving force for both practical and theoretical conquests. So, looking at the developments within the mineral domain is like looking at the general evolution of expert systems itself.

At the turn of the decade, the picture became less crisp. Many more initiatives surfaced. For management of natural resources, incipient work in areas such as forestry left significant traces. This can be seen in conjunction with the efforts initiated around agricultural problems. Problems related to farming received a boost in interest at the end of the 1980s and have contributed to a number of expert systems. Systems dealing with forestry, water, and the like tend to blend into purely agricultural problems. Crop cultivation and forestry share some of the same problems. Pest management is an example of this. When delimiting the scope of this discourse, we have tried to keep out expert systems that are predominantly agriculture oriented. These types of systems deserve to be addressed separately. Yet it is often hard to draw a borderline between the domains.

Ecological systems have emerged as archetypes of complex knowledge-based systems. Often dealing with difficult data fusion supported by remote sensing, they cover aspects of both local and global value. The recognition that many ecosystems are fragile and vulnerable to unexpected influences has set forth activities aimed a wildlife preservation and habitat conservation. Vegetation and atmospheric condition monitoring are also important. The domain of ecosystem management is rich in different types of expertise and demands. The common focus for most of them are the state and health of the environment. Other ES types to be concerned with are those dealing with pollution and accidental releases from chemical plants and similar. However, the objective of most of these is the optimal safety and reliability of the plant. Other systems are more medically inclined and focus mostly on human health.

The selection of systems presented here reflects the general development of applications within the field of management of natural resources. The novel features that each system pioneered or emphasized are highlighted. The novelties of the early systems often became the "bread and butter" of many of those that came later. Understanding how the early, simple ones work will help us comprehend the fundamental structure of the more complex systems.

Diversity in terms of technology is a key word. Systems representing different approaches have been deliberately focused on. This enables both comparison and a better overview. Both prototypes and commercial successes have been addressed. Accounts on prototypes tend to focus on technical aspects only, while reports on deployed systems and systems for sale address usability features. Both are relevant and interesting. The blend presented here is meant to expose elements of importance to both system development, system deployment, and commercialization.


Previous Table of Contents Next

footer nav
Use of this site is subject certain Terms & Conditions.
Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.