Fast, Accurate Creation of Data Validation Formats by End-User Developers

Inputs to web forms often contain typos or other errors. However, existing web form design tools require end-user developers to write regular expressions (“regexps”) or even scripts to validate inputs, which is slow and errorprone because of the poor match between common data types and the regexp notation. We present a new technique enabling end-user developers to describe data as a series of constrained parts, and we have incorporated our technique into a prototype tool. Using this tool, end-user developers can create validation code more quickly and accurately than with existing techniques, finding 90% of invalid inputs in a lab study. This study and our evaluation of the technique’s generality have motivated several tool improvements, which we have implemented and now evaluate using the Cognitive Dimensions framework.