Predict User In-World Activity via Integration of Map Query and Mobility Trace

People often resort to map search engine or other locationbased services for location information when planning long trips or local navigation, and their map queries as well as mobility trace will be accumulated and stored in user log. These data offers valuable information for studying the mechanism of human mobility pattern, furthermore, map query data enable us to sense users’ real-time interests towards locations, and even to forecast their in-world activity in the near future. In this paper, we unveil the connection between users’ map queries and their in-world explorations, and prove the predictability of query-activity formation using two large-scale datasets, the complete map query log and mobility traces of 4 million Baidu map users, which comprise of 118 million map queries and 6.5 billion GPS location records during consecutive 3 months. To the best of our knowledge, it is the first attempt to extensively assess the unique qualities of map query data and predict whether queries about one location would actually lead to in-world visits using heterogeneous data sources. We first characterize the properties of these two datasets, then extract interesting features to quantify their correlation, finally we construct gradient boosting model for prediction, and describe applications empowered by our findings, such as mobility modeling and urban flow estimation.