OSMNames - Local Open Street Name Database

There is a need for a data set consisting of street names (geo names) of the world. Such gazetteer data however is either not available for every country (openaddresses.io) or is not in a suitable format. Furthermore, if such data is found, it is often not for free.

Another problem we face is formatting postal addresses. Each country has its own set of formatting rules and in order to have a database of addresses one needs to bring these different formats to a common format and vice versa.

This thesis includes work in the following areas: First and foremost, the main objective was creating a global and structured data set with street names with the help of OSM data. Additionally, similar products, mostly Nominatim had been analyzed in order to get familiar with the topic. Second, the workflow had to be dockerized so that it can be easily deployed on any system. Finally, an export in a suitable format had to be made publicly available so it can be used in conjunction with a geocoder.

A data set containing planet data could be provided. A secondary target, importing additional house numbers for each street, could not be met in scope of this project. The workflow could be setup in such a way, that it can easily be setup anywhere (via Docker). The simplicity of the installation and, most of all, the clear arrangement of code is a big advantage over other products like Nominatim. After all, the processing run times are way faster than the latter, which takes up to several days for a global data set.

The resulting and easy accessible data export has been successfully integrated by
Klokan Technologies with a SphinxSearch powered geocoder.