Pattern and Phonetic Based Street Name Misspelling Correction

With increasing information reproduced each year, the quality of the data is subject to damage. Variations and errors in the address fields of databases make address-matching strategies problematic. In this paper, first the sources of the error in street names are investigated. Then, a hybrid approach, which is a combination of pattern and phonetic-matching algorithms, is used to fix the problems in the street names. The algorithm is evaluated in terms of its efficiency and correction rates. The newly introduced technique shows an improvement of at least 7% in correction rates compared to the well-known spelling correction algorithms.