Personal name identification through phonetic codification systems

Abstract

The need to identify the variants of personal names is a well-known problem in applications such as information retrieval systems (IRS), digital libraries, databases of patients in a hospital, the electronic systems of air reserves, or the systems of census. The phonetic codification methods constitute one of the procedures for the solution of this problem, permitting to obtain canonical or normalized names. These systems are included inside the general techniques of approximate string matching. In this work a revision of the processes is carried out that utilize the Soundex, Daitch-Mokotoff Soundex, Phonix, Metaphone and NYSIIS systems for the assignment of phonetic keys. The phonetic codification permits reduce to a common form those personal names that are similar in its pronunciation; performance simpler the string matching due to that the common code is stored instead of the complete name. Nevertheless, these systems are dependent of the language utilized, doing necessary the execution of modifications according to the language on the one that apply.