The long(itude) and the short on mapsOctober 8th, 2007
First up, a confession. I like maps.
There, I've said it. But don't change the channel just yet, this just might get interesting. Maps have been long, some might say unjustly, relegated to the too hard basket in libraries, a point driven home by Waikato University's Map Librarian John Robson in a presentation at LIANZA titled Shhh! - if we don't mention maps, they might go away. I'm not going to go into the rights & wrongs, he said/she said of that here, but rather try and articulate my thinking around using maps and geographic data as another entry point into our digital collections.
TAPUHI, our database of unpublished material in the Alexander Turnbull Library, has authority records for places and geographic features. If we can plot this data on a map, or series of maps, we should be able to come up with a nice visual, geographic browse for all of the images on Timeframes. As I see it (and please correct me if I'm mistaken), there are 2 ways we can do this:
- the kinda hard & clunky way &
- the slightly less hard more inclusive, social way.
Catchy, huh? I'm not asking you to choose, though, as I'm not proposing that one way is better than the other. I am in fact proposing that we do both!
One Way: The kinda hard & clunky
How do we get geodata into, or out of the records that exist in Timeframes? It just so happens that Land Information New Zealand publish a downloadable dataset of delimited text tables which have been derived from the New Zealand Geographic Place Names database. This dataset contains information such as Place Name, Northing & Easting from the Universal Transverse Mercator coordinate system , Land District Code, and Latitude and Longitude.
We could do this by matching the place name data from the LINZ dataset with the Turnbull Library's Geographic Name Authority list (whose data has come from the LINZ database/gazetteer) and we should get Latitude & Longitude data for place names which we can then add to our metadata so we can start representing our digital images on a map. Simple, huh? This actually isn't that hard, it just involves a bit of data crunching, and then adding the latitude & longitude to the metadata. Then building the map.
Another Way: The slightly less hard more inclusive, social way
Let the people help! There are a fair few sites around that enable users to add data to their 'own' maps online. For example Zoomin, Yahoo via Flickr and of course, Google Maps. Anyone can create their own maps, and just to prove it, here's one I made earlier adding some Timeframes images to a Google map of Ngaio.
We don't need to harvest the source of these pages to get all the geospatial data. Along with the links to print/save etc on the map there is an intriguing option KML, which just happens to stand for Keyhole Markup Language, an XML-based language for managing the display of three-dimensional geospatial data in the programs Google Earth, Google Maps,Google Mobile and WorldWind. In other words, Latitude and Longitude and stuff.
<description>New York City</description>
<name>New York City</name>
The really interesting to think about is the option of having people create their own maps & sending us the KML, en-masse, rather than us trying to geocode the images from the place authority (which would result in hundreds of images stuck on one spot in Ngaio, for instance). Apparently some folk are already making such maps IRL so there is a place to start from - it just needs some thought and co-ordination.
Things to think about:
In general, the metadata attached to an image on Timeframes records the location of the view, not where the photo was taken from. If we are geotagging where the photo was taken, can we/should we not also get data relating to the view? If we could search on views of specific objects eg Mt Taranaki, as opposed to where the photo was taken from eg Stratford, it may be better for search/retrieval of accurate results.
This may also be too hard, or too labour intensive. I'd be interested to hear what you think.
Interesting things are being done with geographic information. Rheinhold Behringer vhas created an EXIF Editor/GeoTagging tool (No longer up - Ed) for Windows. This integrates with Google Maps for geotagging (scroll down near the bottom of the page) & is also capturing the field of view of the camera, based on parameters stored in the EXIF data about the camera chip and the focal length. In addition to the object coordinates, the camera viewing direction is computed and stored in the EXIF data explicitly.
He does note, however:
Many programs and website support the geo-tagging of the camera position. However, it is not known if any software supports the encoding of the object location. Therefore, this feature must be considered experimental and unique to this particular EXIF GeoTagger software.
Definitely some potential there, but strictly for images taken with digital cameras, not much help for our digitised photos. But, as more and more of our collections will be made up of born digital images, this is definitely worth keeping an eye on.
This has been a bit of a ramble, and I'm still trying to solidify my thoughts - there's so much going on. With the expansion & development of map APIs, there are more and more ways we can provide access to our collections. Now you can even map a plate of pasta! (No longer up - Ed)
Has anyone else out there done some thinking on this that could help me out? Your comments and suggestions are most welcome.