OMS Builds Nationwide Building DB and Internalizes Address Refinement Based on Ministry of the Interior and Safety Navigation DB

To reduce dependency on external APIs and directly develop address refinement functionality, the OMS team built Building Address Refinement System 2.0 based on the Ministry of the Interior and Safety navigation DB. Through about 1.5 months of work, they successfully loaded detailed information for 10,778,620 buildings nationwide into their own DB.

Data Source and Development Strategy

In the early stage of development, the team first processed data from Sejong Special Self-Governing City, the Ministry of the Interior and Safety's headquarters region, to determine whether any data was missing. After verifying about 56,000 building records in Sejong, they proceeded to build the nationwide DB in the order of Seoul, Daejeon, Daegu, and Busan. They developed an API that directly parses the .txt format raw data provided by the Ministry of the Interior and Safety and inserts it into the DB, which included logic to convert GRS80 UTM-K coordinates into geographic latitude/longitude.

Technical Trial and Error, and Lessons Learned

During the data loading process, trying to extract only the necessary fields led to the inefficiency of having to go back and fill in missing fields. This led to the lesson that brute-force inserting all columns first and then determining necessity afterward is better in terms of both mental health and efficiency. For example, the 'underground' status value initially seemed unnecessary, but was confirmed to be essential for handling special cases such as underground addresses in Sasang-gu, Busan.

Operational Automation and Synergy Effects

Through cooperation with a Ministry of the Interior and Safety contact, the team secured a private daily sync API that automatically updates nationwide building changes every morning at 6 AM. This allows information such as newly occupied apartment complexes to be reflected in real time. The biggest synergy from building the building DB is the improvement of the OMS deliverable-area cache structure. This overcame the limitations of the existing postal-code-based cache management, using the building management number as a key to reduce memory usage and ensure cache consistency.