Imagery Derived Attributes for Activity Based Intelligence
Anonym
In a few weeks I'll be attending the 2012 Esri International Users Conference and I'm really looking forward to an Intelligence Community session that features the emerging discipline called Activity Based Intelligence (ABI). In a broad sense, ABI is defined as the observation (a.k.a., collection) and analysis of the actions of individuals or groups as they interact with each other and their environment. Like I said, broad, but as such ABI is a fitting topic for a conference full of Geographers.
The activities of individuals and groups are easily observed using the latest advances in imaging technologies like overhead persistent surveillance and full spectrum remote sensing (e.g., visible, IR, SAR, etc.). These sensors immediately put activity observations in context with their physical environment, but they provide little insight into human elements like the motivation for an activity or how people are reacting to the activity.
Enter social media.
ABI uses social media as a collection source to add the human dimension to the analysis. After all, someone, somewhere, is likely to be posting something from their perspective about an activity. The something may be the most valuable piece, but it's the somewhere that intrigues me most.
The somewhere is a place or geolocation that links the human element, captured in the social media, to the physical element, captured by our sophisticated imaging sensors. And there’s a lot to be gained by making this link. Image analysis can provide information on the type of terrain, land cover, view obstructions, exposure, and weather that exists where the social media source is immersed. Tying these types of imagery derived attributes to the social media post could be used to establish the veracity of the post or social media source in general.
For example, while synthesizing eye-witness reports on violence in Syria, an analyst may give more weight to a post from an urban environment than to a post from a rural environment. Or using imagery analysis an analyst could establish if an event was viewable from the location of the post. Was there a line of sight or should the post be discounted because there were too many obstructions from the vantage point of the posts author?
Adding attributes derived from imagery could also be useful in establishing a location pattern or an environmental profile for individual posters. For example, ABI analysts might only care to analyze posts from individuals who consistently post from an environment of interest. A social media source may be deemed more credible if they have been posting from that environment for a long time.
In a few weeks, I'll be posting from the Esri UC in beautiful San Diego, CA. What if my boss checks the physical environment of my post locations? Attributes like urban, city center, concrete, and within 50m of the convention center would bode well for my performance review, but wouldn't beach, sand, palm trees, and within 50m of the Pacific Ocean be better for my tan?