Heuristic Search

Abstract

This text has focused on the interaction of algorithms with data structures. Many of the algorithms presented in this text deal with search and how to organize data so searching can be done efficiently. Many problems involve searching for an answer among many possible solutions, not all of which are correct. Sometimes, there are so many possibilities, no algorithm can be written that will efficiently find a correct solution amongst all the possible solutions. In these cases, we may be able to use a rule of thumb, most often called a heuristic in computer science, to eliminate some of these possibilities from our search space. If the heuristic does not eliminate possible solutions, it may at least help us order the possible solutions so we look at better possible solutions first, whatever better might mean. In Chap. 7 depth first search of a graph was presented. Sometimes search spaces for graphs or other problems grow to such an enormous size, it is impossible to blindly search for a goal node. This is where a heuristic can come in handy. This chapter uses searching a maze, which is really just a type of graph, as an example to illustrate several search algorithms that are related to depth first or breadth first search. Several applications of these search algorithms are also presented or discussed. Heuristic search is often covered in texts on Artificial Intelligence [Cop04]. As problems in AI are better understood, algorithms arise that become more commonplace over time. The heuristic algorithms presented in this chapter are covered in more detail in an AI text, but as data sizes grow, heuristic search will become more and more necessary in all sorts of applications. AI techniques may be useful in many search problems and so are covered in this chapter to provide an introduction to search algorithms designed to deal with large or infinite search spaces.