When (not) to Use Derivational Analogy: Lessons Learned Using APU

Sanjay Bhansali

We present an experience report in applying derivational analogy to speed up the performance of a prototype system called APU that synthesizes UNIX shell scripts from a problem specification. We present experimental results showing how the system performed, the quality of solutions obtained by using analogy, and how the system scaled up with size of case libraries. We discuss the implications of the results and describe properties of the knowledge representation and the domain that contributed to APU’s performance.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.