Deriving human activity from geo-located data by ontological and statistical reasoning

Abstract

Every day, billions of geo-referenced data (e.g., mobile phone data records, geo-tagged social media, gps records, etc.) are generated by user activities. Such data provides inspiring insights about human activities and behaviors, the discovery of which is important in a variety of domains such as social and economic development, urban planning, and health prevention. The major challenge in those areas is that interpreting such a big stream of data requires a deep understanding of context where each activity occurs. In this study, we use a geographical information data, OpenStreetMap (OSM) to enrich such context with possible knowledge. We build a combined logical and statistical reasoning model for inferring human activities in qualitative terms in a given context. An extensive validation of the model is performed using separate data-sources in two different cities. The experimental study shows that the model is proven to be effective with a certain accuracy for predicting the context of human activity in mobile phone data records.