Empirical Study in the Simulation of Heuristic Error Behavior

摘要

Many artificial intelligence programs deal in searching for solutions from many alternatives, and depend upon heuristics to guide the search and to insure quality results. This thesis documents empirical research performed in searching problems and heuristic behavior. In this thesis a general model is presented that defines a broad class of searching domains, along with a set of software tools designed to support research in them. One puzzle configuration is devised from the model and studied in depth, examining several subtle variations of the A* searching algorithm, and results are compared with related work from other similar domains. (Author)