IP Geolocation through Reverse DNS

IP Geolocation databases are widely used in online services to map end-user addresses their geographical location. However, they use proprietary geolocation methods, and some cases have poor accuracy. We propose a systematic approach reverse DNS hostnames for geolocating addresses, with focus on as opposed router IPs. Our method is designed be combined other data sources. cast the task machine learning problem where, given hostname, we first generate list of potential location candidates, then classify each hostname candidate pair using binary classifier determine which candidates plausible. Finally, rank remaining by confidence (class probability) break ties population count. evaluate our against three state-of-the-art academic baselines two commercial databases. show that work significantly outperforms complementary competitive To aid reproducibility, open source entire make it available community.