Quality evaluation of postal address datasets measuring their autocorrelation

Abstract

Many spatial applications related to land and titles like land use management, registering, and utility and health service providers are using postal addresses as their main or their supplementary georeferencing method. Evaluation of postal address datasets quality is important when controlling their changes due to manipulations (like add or update), comparing them, or merging them, that is one of the main strategies of developing countries like Iran, to form a unified addressing structure and database. Despite the costly and time consuming formal methods of postal addresses qualification that are based on address matching, the method proposed in this paper provides an evaluation of a postal address quality not requiring any preprocessing like standardization or ancillary data like streets and their addressing scheme data. The proposed method is based on measuring the autocorrelation of a postal address dataset content where higher level of autocorrelation indicates more standardization and less spatial sparsity of the addresses. The method processes the adjacency graph formed measuring Damerau–Levenstein distance between records of a postal address dataset. Evaluation of 5 statistics for 4 postal address datasets of Tehran City of Iran shows that the cumulative frequency of values and the maximum size of the components (sub-graphs) in the adjacency graph could be used. These statistics both show stable S-Shaped patterns that their threshold at the first extremum of their second derivative represents the desired quality of a postal address dataset. The results show that the measured threshold of postal address dataset corresponds with its topological structure of the streets that cover its addresses. The method can define characteristics of a standard address structure for one or more postal address datasets as the results propose 5 components for the standard address of the evaluated datasets which is the same as the number of components defined for Iranian national structure of postal addresses.