Next: Problem Assumption 1: To
Up: Assumptions of Direct Comparison
Previous: Other Applications
General-purpose planners exhibit differential capabilities on domains
and sometimes even problems within a domain. Thus, the selection of
problem set would seem to be critical to evaluation. For example, many
problems in benchmark sets are variants of logistics problems; thus, a
general-purpose planner that was actually tailored for logistics may
appear to be better overall on current benchmarks. In this section, we will
empirically examine some possible problem set factors that may
influence performance results.
©2002 AI Access Foundation and Morgan Kaufmann
Publishers. All rights reserved.