EXAMININIG THE SIMILARITY OF POSTAL ADDRESS COMPONENTS IN TURKEY

Abstract

Studies on standardization of addressing system have been conducted by the regional and international organizations in many countries. However, they have not received much attention in countries such as Turkey where the address systems have been newly formed and developed. The objective of this study is to examine the component-driven similarities of address data used in Turkey after parsing the postal addresses into meaningful components. In this context, the postal addresses of the accommodation buildings in the district of Fatih in Istanbul are analyzed using the Damerau-Levenshtein distance algorithm, which is known as one of the string similarity measures in binary. These addresses are collected by using web scraping method from five different web portals and obtained by geocoding from four different map services. The reference data for the comparison process is taken from the Turkish Address Registration System (TARS). The results show that each component does not have similar accuracy.