The paper studies the Optimal Decision Tree (ODT) problem, in which a decision tree must identify an unknown hypothesis from a given ground set. The cost of the tree is measured in either the average case or the worst case, and the problem has been extensively studied.
The main contribution is a set of logarithmic approximation algorithms that the authors describe as simpler than previous approaches. The same framework also applies to Adaptive Set Cover, a related problem.
Because this summary is based on a single arXiv abstract, there are no independent results to compare against. The significance is the promise of easier-to-use algorithms that still achieve logarithmic approximation guarantees.