An Analysis on Transformational Analogy: General Framework and Complexity

摘要

In this paper we present TransUCP, a general framework for transformational analogy. Using our framework we demonstrate that transformational analogy does not meet a crucial condition for a well-known worst-case complexity scenario, and therefore the results about plan adaptation being computationally harder than planning from the scratch does not apply to transformational analogy. We prove this by constructing a counter-example that does not meet this condition. Furthermore, we perform experiments that demonstrate that this counter-example is not an exception. Rather, our experiments show that it is unlikely that this condition will be met when performing plan adaptation with transformational analogy.