Publication Details
Fuzzy matching as a technique for the name correction
Fuzzy match, name correction, approximate match, soundex, name recognition
This paper introduces methods that can be used to correct those errors in electronic mail addresses which occur when a recipient name in the header of an e-mail message is misspelled or mistyped. The target mail delivery system should automatically detect the error and correct it. We use the term "fuzzy match" to mean the best approximate match in this context. By combination of various techniques -- simple string comparison, approximate matching up to 2 allowed mistakes, soundex matching etc. -- we implement fuzzy matching to achieve good results in a reasonable time. This paper covers an analysis of applied techniques and their drawbacks (i.e. problems with search speed, search imprecision and misinterpretation). It also gives some ideas for the future improvement. Following this strategy a program called mailrouter was designed which demonstrates practical employment with required results. The project was realized at CERN, Geneva, Switzerland as a part of technical student programme.
@inproceedings{BUT191832,
author="Petr {Matoušek}",
title="Fuzzy matching as a technique for the name correction",
booktitle="MOSIS '99, Proceedings",
year="1999",
volume="Volume 2",
pages="85--91",
publisher="unknown",
address="Rožnov p.R.",
isbn="80-85988-33-X",
url="http://www.fit.vutbr.cz/~matousp/doc/1999/mailrout.ps"
}