Cluster-based Find & Replace

Alisa Marshall & Robert Miller

Abstract
In current text editors, the find & replace command offers only two options: replace one match at a time prompting for confirmation, or replace all matches at once without any con- firmation. Both approaches are prone to errors. This paper explores a third way: cluster-based find & replace, in which the matches are clustered by similarity and whole clusters can be replaced at once. We hypothesized that cluster-based find & replace would make find & replace tasks both faster and more accurate, but initial user studies suggest that clustering may improve speed on some tasks but not accuracy. Users also prefer using a perfect-selection strategy for find & replace, rather than an interleaved decision-action strategy.

Cluster-based find & replace in action
References:

[1] Robert C. Miller and Alisa A. Marshall. "Cluster-based Find & Replace." Conference on Human Factors in Computing Systems (CHI 2004), April 2004.

[2] Alisa Marshall. "Cluster-Based Find & Replace." MEng thesis, Massachusetts Institute of Technology, June 2003.