[First Hit]

Datapages, Inc.Print this page

ABSTRACT: Sedimentological Previous HitInterpretationNext Hit Using Expert Systems

MCLEAN, J. ROSS, Consultant, Calgary, Alberta, Canada

Expert systems have a strong intellectual appeal, but they often fall short of expectations. There are four major areas of concern in the use of an expert system for sedimentological Previous HitinterpretationNext Hit: (1) the great complexity and variability in nature, (2) organization of data into a viable computer program, (3) designing for use by a geologist inexperienced in sedimentology, and (4) designing for use by an experienced sedimentologist.

The first issue can be addressed by starting at a high level of Previous HitinterpretationNext Hit with depositional systems and eliminating possibilities until only one viable Previous HitinterpretationTop remains. Data, in the form of attributes, can be arranged in three manners: as rule-based, frame-based, or matrix-based systems. For a sedimentological expert system, the latter is favored.

A great difficulty in effective use of the expert system is that the less experienced geologist will not be able to recognize the attributes necessary to allow the system to work effectively. The expert system can be set up to try to overcome this obstacle in three ways: (1) a user/computer interactive interface, (2) a probability matrix that scores the likelihood of two environments being juxtaposed vertically, and (3) development of an extensive educational function that allows the user to easily access an illustrated library of definitions and examples.

Experienced sedimentologists may feel they do not need such a system. However, there are two enticements to using it: (1) a high-quality graphical output that makes the package attractive by itself, even without the expert system, and (2) the identification of multiple possibilities to overcome personal biases. Such a system is not only highly useful, but saves time and forces the user to be systematic.

 

AAPG Search and Discovery Article #91012©1992 AAPG Annual Meeting, Calgary, Alberta, Canada, June 22-25, 1992 (2009)