Can Agree and Labeling Be Reduced to Minimal Search?

Abstract

This article develops a formal definition of Minimal Search to evaluate the idea that Agree and Labeling can be reduced to Minimal Search. Different aspects of the search algorithm used in Minimal Search (i.e., breadth-first vs. depth-first search, parallel vs. serial search, global vs. modular search) are compared, and reasons for choosing between each of these pairs are given based on detailed examinations of their theoretical and empirical consequences. Given the formal definition of Minimal Search, the article argues that Agree and Labeling can only be partially unified by Minimal Search: the search algorithms used in Agree and Labeling can be unified by Minimal Search, but the values of the search targets and search domains are determined by Agree and Labeling independently. On the basis of the definition of Minimal Search, the article (re)defines Agree and Labeling to capture both the similarities and the differences between these two operations.