Abstract: Successful intelligence critically relies on meaningful, correct, and quick interpretation of patterns gathered from data. While the quantity of available raw geospatial data is overwhelming for a human analyst, statistical and machine learning methods for pattern identification often lack necessary context about the relevance of observed features and activities as they develop through time. In this talk, I will outline a possible path for bridging the gap between context and computation. The work relies on exploiting the domain-specific insight of an experienced intelligence analyst to direct a computer to automatically identify those features and patterns that have intelligence value, whether to describe the community patterns of life or to identify single actors of interest. The key element of the approach is temporal logic: a language for describing observed patterns over time in a way that is both understandable to a human and amenable to machine reasoning. Our ongoing work seeks to build a pipeline from the analysts insight, through information-theoretical notions that encode the importance of particular features, to algorithmic identification of observed activities and behaviors that may be of intelligence interest. To provide a proof of concept, I will describe some preliminary results on learning driver and rider behavior from data on trips taken by taxi and rideshare services in the US.