Public checkins versus private queries: measuring and evaluating spatial preference

Understanding the spatial preference of mobile and web users is of great significance to creating and improving location-based recommendation systems, travel planners, search engines, and other emerging mobile applications. However, traditional sources of spatial preference -- which reflect the patterns of geo-spatial interest of large numbers of users -- have typically been expensive to collect, proprietary, and unavailable for widespread use. In this paper, we investigate the viability of new publicly-available geospatial information to capture spatial preference. Concretely, we compare a set of 35 million publicly shared check-ins voluntarily generated by users of a popular location sharing service with a set of over 400 million private query logs recorded by a commercial hotel search engine. Although generated by users with fundamentally different intentions, we find common conclusions may be drawn from both data sources -- (i) that the relative geo-spatial "footprint" of different locations is surprisingly consistent across both; (ii) that methods to identify significant locations results in similar conclusions; and (iii) that similar performance may be achieved for automatically identifying groups of related locations. These results indicate the viability of publicly shared location information to complement (and replace, in some cases), privately held location information.