Representing and reasoning about change (abstract only)

A recent trend in artificial intelligence research is the construction of expert systems capable of reasoning from a detailed model of the objects in their domain and the processes that affect those objects. We describe a system being built in this fashion, designed to solve a class of problems know...

Full description

Saved in:
Bibliographic Details
Published in:Computer graphics (New York, N.Y.) N.Y.), 1984-01, Vol.18 (1), p.27-27
Main Authors: Simmons, Reid G., Davis, Randall
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A recent trend in artificial intelligence research is the construction of expert systems capable of reasoning from a detailed model of the objects in their domain and the processes that affect those objects. We describe a system being built in this fashion, designed to solve a class of problems known as geologic interpretation: given a cross-section of the Earth's crust (showing formations, faults, intrusions, etc.), hypothesize a sequence of geologic events whose occurrence could have formed that region. Solving the geologic interpretation problem requires reasoning about change, in particular, spatial change. The shape of a formation, for example, can be altered by the process of erosion. Doing this reasoning, in turn, requires representing objects, which show the effects of change, and processes, which are the causes of those changes.The main focus of this research is to explore the machinery needed to represent and reason about both mutable objects and the processes that induce changes in them. To do this, we have developed two representations of objects, one involving histories and the other involving diagrams . We have also developed two corresponding representations of physical processes, each suited to reasoning about one of the object representations. We have been careful to keep the two representations well separated, limiting their interaction to a relatively small and dearly defined interface.We have used these representations to model a subset of geology large enough to allow us to solve most geologic interpretation problems. In particular, the model allows us to imagine a sequence of geologic events. Imagining is a new form of qualitative simulation similar to envisioning .
ISSN:0097-8930
DOI:10.1145/988525.988553