Nonmonotonic Extrapolation of Causal Relations for Knowledge-Based Decision Support Using a Bayesian Network Approach
Abstract
This research focuses on an investigation of a new computational model for providing reliable decision support in military operations. The PT used a Bayesian network representation and a human-centered reasoning technique to extrapolate the causal relations between the available data sets stored in heterogeneous databases and the phenomena implications coherent to real world situations. The PT had developed a software agent interface for integrating the data mining and decision support operations. A nonmonotonic reasoning paradigm for derivation of causal relations was implemented in an integration of relevant software modules. The agent interface was featured with an interactive graphics setting for display and manipulation of, the Bayesian Network representations, for multiple database accesses, and for belief propagation. An interactive and iterative knowledge acquisition and Bayesian reasoning system prototype was developed on top of the agent interface.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2002
- Accession Number
- ADA409265
Entities
People
- Qiuming Zhu
Organizations
- University of Nebraska Omaha