Scheduling consists mainly of allocating resources to jobs over time under
necessary constraints. In the past, the processing time for each job was us
ually assigned or estimated as a fixed value. In many real-world applicatio
ns, however, job processing times may vary dynamically. McCahon and Lee pro
posed a fuzzy Johnson algorithm for managing uncertain scheduling. However,
some problems exist in their procedure with respect to calculation of the
starting time for each job. In this paper, we modify McCahon and Lee's algo
rithm and propose a new reasonable procedure for eliminating start-time unc
ertainties. A half-inverse operator is defined and 24 cases are analyzed to
verify this procedure. Analytical and experimental results showing the eff
ectiveness of our method are also presented. (C) 1999 Elsevier Science Ltd.
All rights reserved.