“…Measurement selection requires a measurement selection function [3,91,92] which gets a set of minimal diagnoses D as input, and outputs one system measurement such that any measurement outcome eliminates at least one spurious diagnosis in D. As measurement selection functions we adopted split-in-half (SPL) [32], which suggests a measurement with the lowest worst-case number of spurious diagnoses in D eliminated 44 , and entropy (ENT) [3], which selects a measurement with highest information gain. These functions appear to be the most commonly adopted ones in model-based diagnosis, cf., e.g., [18,32,92,93,94,95,96,97,98].…”