After watching the Microsoft Virtual Earth spiel at their CEO summit (http://www.microsoft.com/winme/0605/27736/BillG_CEO_Summit_MBR.asx )earlier this year it reinforced that the geospatial web has still not really gotten past just putting push pins on maps. Don’t get me wrong MS, Google, and Yahoo and the various mash ups they have inspired have done some incredibly cool stuff with putting push pins on maps, but it has not yet evolved to providing true analysis of the push pins that allow users to make better decisions.

The one place where you do see analysis going on is with driving directions, but even that is really just starting to evolve past what Mapquest did years and years ago. In my mind the real contribution of the geospatial web to date has been unleashing the huge amount of geo-referencable data that has been sitting dormant. The easy to use features of Google Earth and KML really kicked it off by providing a dynamic and cool mapping widget for people to look at theirs and others data. The result was a huge number of mashups many times with data no one had seen on a map before like locations of houses for sales with pictures and prices or the location of registered sex offenders by street address.

In addition to looking at the push pins of where sex offenders are and where houses for sale are, the consumer should be able to numerically compare the concentration of sex offenders and home prices in one location versus another location. Go the next step and add in schools, test scores, and user rated Mexican restaurants. What are the locations that have the highest concentration of amenities to make a location attractive to home buyer, business location, marketing campaign, franchise expansion, warehouse etc. You could also calculate the concentration of risk to natural hazards for insurance and security uses. Once you start creating geo-analytics that are easy and intuitive to understand to non-technical users there are a whole host of questions people can start to answer with their push pins on maps.

This is where there is a considerable gulf between the geospatial web and traditional GIS. The geospatial web has made mapping technologies available to the masses but has not been able to provide analytics. Traditional GIS has a vast array of analytics but they are so arcane and technical only formally trained professional can use them. The trick is to harness the analytics of traditional GIS into the easy to use world of the geospatial web. Two big problems block the road to what seems like a straight forward trip. For one the average person has no clue what traditional GIS analytics are or how to perform them – in fact few professionals really understand the mathematics behind what is being done. The intricacies of inverses distance interpolation or Gaussian decays of kernel density functions are lost on 99% of the universe. Yet these are exactly the tools needed to answer the simple user question discussed above.

That is the first bridge to be crossed, but even if you do manage to simplify the labyrinth world of map algebra, you still have severe computational limitation to surmount if you want to deliver analytics to a browser. If you have every tried to run a kernel density analysis with a decent pixel resolution across a large geography on a traditional GIS – you might as well make a coffee run because it is going to be a while before you get a pretty heat map back. This is using a desk top application, not sending it to a browser, and it is creating only one heat map (raster surface). Since Google Earth the mass users are used to getting more detail when they into an image, and they will expect the same of their analytics. Producing raster heat maps on the fly is something that even high powered desktop applications cannot achieve.

We think there are creative ways to solve these problems, and hope to start a discussion with community as we get ready to launch our approach to it. Creating and any and all feedback or ideas about how geo-analytics can be evolved to create value for the geospatial web is the goal of the blog. So join in and hopefully we’ll have something cool to show in the next month or so.

Thanks,

Sean Gorman

Popularity: 17% [?]