An address geocoding solution for Chinese cities

摘要

We introduce the challenges of address geocoding for Chinese cities and present a potential solution along with a prototype system that deal with these challenges by combining and extending current geocoding solutions developed for Unite States and Japan. The proposed solution starts by separating city addresses into "standard" addresses which meet a predefined address model and non-standard ones. The standard addresses are stored in a structured relational database in their normalized forms, while a selected portion of the non-standard addresses are stored as aliases to the standard addresses. An in-memory address index is then constructed from the address database and serves as the basis for real-time address matching. Test results were obtained from two trials conducted in the city Beijing. On average 80% matching rate were achieved. Possible improvements to the current design are also discussed.