Evaluation of Uncertain Inference Models I: PROSPECTOR

TL;DR

This paper evaluates the accuracy of the PROSPECTOR uncertain inference model by comparing its solutions to probability theory and cross-entropy calculations across various inference networks.

Contribution

It provides a systematic assessment of PROSPECTOR's accuracy and identifies conditions affecting its performance within its assumed problem subset.

Findings

PROSPECTOR is generally accurate within its assumptions

Performance deteriorates under certain conditions

Comparison with probability theory highlights strengths and limitations

Abstract

This paper examines the accuracy of the PROSPECTOR model for uncertain reasoning. PROSPECTOR's solutions for a large number of computer-generated inference networks were compared to those obtained from probability theory and minimum cross-entropy calculations. PROSPECTOR's answers were generally accurate for a restricted subset of problems that are consistent with its assumptions. However, even within this subset, we identified conditions under which PROSPECTOR's performance deteriorates.