Mining geographic knowledge using location aware topic model
Chong Wang, Jinggang Wang, Xing Xie, Wei‐Ying Ma · 2007
Most online activities are associated with geographical locations. For example, people write personal blogs about interesting places they have ever been to; read news about important local events; and search the web to find delicious restaurants. Mining geographical knowledge from these online activities can greatly benefit lots of web applications. In this paper, we propose a Location Aware Topic Model (LATM), a probabilistic graphical model, to explicitly model the relationships between locations and words. Experiments on several data sets, including news and blogs, showed satisfactory results.