Tim Warr’s Blog

I have been taking some time to explore there latest spatial thinking and innovations by Yahoo!.  Setting aside some of the current business politics, I think Yahoo! are doing some of the most interesting stuff at the moment. 

It started a while ago with their Where On Earth (WOE) IDs.  Basically these provide a hierarchical identifier system based on location.  I must admit it took me ages to realise what these are all about.  I have a map centric view centric view of the world and places, so it seemed very odd at first.  WOE IDS geo-tag (i.e. locate) things in a very different way (with no need for a map or lat/lon).  When talking about a place all people really care about is what place it is and how to ensure that they are all talking about the same place (even if you might individually define its extent differently).  Lets take an example, if we want to share data relating to Soho, how do we make sure we are talking about the same place?  So with this system we do not have to argue endlessly about what the lat/lon of Soho is (or even worse try and define it’s extent), all we have to agree on is that Soho is in London which is in England which is in Great Britain, give it an ID and then we know we are talking about the same place… 

So what I love about WOE ID is it is a simple non mapping solution to an age old spatial problem.  In fact a map is ultimately relegated to just a display function.

So ahy has this now got really interesting?  Well Flickr has been storing WOE IDs for geo-tagged photos.  From this Yahoo! have been able to predict the extent of places (with WOE IDs) based on the geo-tags for the relevant Flickr photos.  The resulting geometries are know as Alpha Shapes.  So this is a completely different way to discover more about the geo-extent of a place.  Taking the Soho example (which BTW is not an officially defined area) we can define it’s extent by what most people (taking photos) think it is.  There are examples and details of this approach here on the Flickr development blog.. 

What I love about this is that, by taking a non map-centric view of place, Yahoo have developed an approach that can tell us more about what the the perceived geographic extent of a place is.  No GIS person would ever think of doing it this way round!

Even better you can get the data yourself (as ESRI shapefiles) and try the software Yahoo use to create the Alpha Shapes (called Clustr).

More details on Alpha Shapes are blogged here too.

Lets hope this innovation continues!