Open access
Aug 2026
On Behavioral Alignment of Model-Code and Human-Code Understandability via Behavioral Proxies
It is demonstrated that semantic self-consistency is a reliable and extensible measure to be used as a behavioral proxy for quantifying model code understandability, with broad implications in both software engineering research and practice.
Xiao-Kai Rong, Aashish Yadavally, Hridya Dhulipala et al.
· Proceedings of the ACM on So... · 0 citations