Location, Location, Location

Spatial data from global positioning and imaging systems is fueling new location intelligence apps.

In today's wired world, where it's at is simultaneously irrelevant—Bangalore and Baltimore are next-door neighbors on the Internet—and a key issue for those who understand the lift that location intelligence can give to traditional analytics.

Location intelligence combines spatial-data collection with analysis and visualization methods that transform siting to knowledge. What's new isn't so much the analysis and visualization components; serviceable geographic information systems (GIS), thematic mapping and routing systems have been available for years. Rather, the most significant recent innovations have been in the data-collection and knowledge-delivery bookends of location intelligence. These innovations should lead to more sophisticated spatial analytics.

We're generating and recording more spatial data than ever with global-positioning system (GPS) enabled devices, via sensor networks that use transponders to track object movements and through systematic Earth imaging. And researchers are repurposing commonly used technologies to generate spatial data, for instance by triangulating wireless networking signals to provide geolocation, service localization and tracking services.

The potential power of geospatial intelligence was demonstrated by a recent diplomatic situation. In June, an Italian magistrate charged 13 Central Intelligence Agency operatives with illegally apprehending and removing a terrorism suspect from Italy after an investigation used cell-phone records to correlate the operatives' movements with the abduction. The flip side occurred when agents followed a decoy Osama bin Laden: An associate reportedly took bin Laden's satellite phone on a road trip in late 2001 while bin Laden headed in a different direction. There are valid concerns about the rules under which geospatial information, like other sensitive data, should be collected, retained and used by government and industry. The growth in the collection and use of geospatial data only reinforces the acuity of Sun CEO Scott McNealy's 1999 dictum, "You have zero privacy anyway; get over it."

Setting aside the legal and privacy concerns, there's some great stuff coming out. Start with Google Earth, an instant-hit application that among other things infers three-dimensional topography from satellite images. It joins Google Maps, also still in beta release, which provides ease of use and an overlay of information from directory databases that bests established Internet mapping sites. While there are other tools of this nature, the key innovation is that Google and also Yahoo! publish APIs that allow anyone on the Internet to disseminate information via navigable map interfaces. A9's Block View, another beta, is another great application, compiled from tens of millions of street-level photographs of U.S. cities to provide location views for services such as's Yellow Pages business locator. The Where 2.0 conference Web site ( is a good place to learn more.

Service locators and thematic maps are the geospatial equivalent of reporting and pivot tables that manipulate multidimensional data cubes. These technologies are great for visualizing information extracted from structured data but can't provide knowledge discovery that can detect—not just display—patterns and relationships. They're also designed for point-in-time viewing and aren't well suited for temporal studies, forecasting or other forms of predictive analysis.

You can use more advanced, model-based location intelligence to decide store and service locations, study population and business dynamics, or detect crime patterns. These models can capture business-domain knowledge and incorporate demographic data and proprietary business information in geographic and numeric form. Maps are only a start; we need algorithms that compute spatial metrics such as distance based on concepts such as density, direction and proximity. The goal is to optimize and predict using geographic and temporal variables in a dynamic and topographically complex world. This type of modeling isn't yet automated, à la data mining, but with the growth in geospatial data, we can't be far off.

Google and Yahoo help us answer "where it's at": Tell me where it'll be next week and how best to get there? Then location intelligence will be in business.

Seth Grimes is a principal of Alta Plana Corp., a Washington, D.C.-based consultancy specializing in large-scale analytic computing systems. Write to him at [email protected].