![]() |
|||
![]()
|
![]() |
![]() |
![]() |
2.7. XX/MAGIC: A WORKSTATION TOOLBOX FOR THE EXPLORATIONISTKendall et al. One of the most pronounced features of this ambitious effort (Kendall 88; Morgan 87) was the fact that it was marketed as an expert database system and a decision support tool. At this point, an expert system was often associated with a single, straightforward rule base of the MYCIN type. Lessons over the past years had taught that knowledge modules of that kind had to be delivered together with other software modules, typically databases and numerical calculation modules in order to provide sufficient benefits for the user. Moreover, the traditional expert system was not associated with the mixed initiative effort. Hence, the decision support label was applied to highlight this and to emphasize the fact that the user would not necessarily have to rely solely on the conclusions made by the program. Another important factor in the commercial world was the term "expert system" itself, which tended to challenge the competence of the user. In other words, using an expert system could diminish the role of the user, something that was unheard of in many situations. XX was the acronym for eXpert eXplorer. It had an extended scope in terms of exploration compared to that of log analysis systems. It supported the identification, modeling, analysis and presentation of potential hydrocarbon plays and prospects. A prospect in terms of exploration is a potential enclosure of hydrocarbons set in a reservoir rock. A play is a geological model at a higher level that includes a number of prospects as well as source rock considerations. The time aspects involving temporal considerations related to maturation and expulsion of hydrocarbons are also significant at this level. A play model also includes migration aspects involving the transportation of the hydrocarbons from the source to the reservoir. In its first version, the XX system included seven main components:
Based on historical data, input from the users, simulations, and its own decisions, XX would produce risk assessments, drilling recommendations, engineering considerations, and various types of graphics: maps, cluster dendograms, charts, etc. The architecture of XX was organized around the cooperation of several knowledge sources similar to that of a blackboard. But in contrast to the standard blackboard, there was no persistent use of a common data exchange area. The control enforced on the different parts of the system was more passive and in the hands of the user. Still, the use of a common database and a central database manager that monitor the functions of the various modules makes XX comparable to that of the META/LOG. Indeed, the communication with the GIS facility, a database, and a knowledge base has emerged as a standard architecture for many systems. Two of the most interesting aspects of XX were the casebook approach and its solution to approximate reasoning. As a case-oriented system, it pioneered in a simple way a paradigm that became extremely popular at a later point in time. The case-based reasoning (CBR) method proved very efficient with the XX system. Surprisingly few records of CBR related to management of natural resources are recorded. Within the domain of geological exploration, the remote sensing application of Bremdal (Bremdal 95) and the petrographic system by Abel (Abel 96), which will be discussed later, are known. In XX, the known characteristics of some predefined plays were matched against the description of the target play. Based on a similarity analysis, a geologic structure was deduced and the hydrocarbon potentials determined. Most of the knowledge captured in XX was in the form of rules like "IF mostly a THEN mostly b." Inexact reasoning involving belief values were successfully applied using the Dempster-Shafer technique. In contrast to PROSPECTOR-type systems, the clause part of a rule is considered a pattern, of which the conclusion part is dependent. There may be several conclusions associated with a pattern, each with a belief factor projecting the expert's degree of belief that the conclusion is related to the pattern on the left-hand side. In the Dempster-Shafer framework, a piece of evidence may support a single hypothesis as a conclusion. It may also support a set of hypotheses as a conclusion. The rule-based modules in XX applying this kind of inexact reasoning were built using a shell called MIDST, Mixed Inferencing Dempster Shafer Tool.
|
![]() |
|
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. |