I recently published a paper that used standard measures of string similiarity
(e.g., bigram, trigram, and edit distance) to predict whether pairs of drug
names were likely to be confused in practice. This paper may be of interest to
some corpora subscribers.
Lambert, B. L. (1997). Predicting look-alike and sound-alike medication errors.
_American Journal of Health-System Pharmacy_, Vol. 54, No. 10, pp. 1161-1171.
A slightly earlier version is available at
ftp://ludwig.pmad.uic.edu/pub/MedErrors.pdf
Comments welcome.
Bruce Lambert
Department of Pharmacy Administration
University of Illinois at Chicago