Allan and I had a productive brainstorming session today, and came up
with a system for organizing Mashable City's source data. Your
feedback would be appreciated!
Ultimately, we want to bring all kinds of local data into structured,
query-able data stores, but to start, we'd like to build an index of
the URLs where this info already lives. For instance, we'd eventually
like to have the full menu, hours of operation, and reviews for 729
Hope Cafe directly accessible, but for now it would help just to have
a record that
http://729hopecafe.com/ has info about something near
the latitude and longitude (41.846383, -71.396283).
Fortunately, we discovered that
http://del.icio.us/ has support for
machine tags (seen also in Flickr), which can be used to store
latitude and longitude info, among other things. More info on this is
available at
http://en.wikipedia.org/wiki/Geotagging#Geotagging_in_tag_based_systems
You can get the lat+lon coordinates for a street address at
http://geocoder.us
Here's a screenshot of how I bookmarked
729hopecafe.com
http://img.skitch.com/20080412-kb2siijyt6by66qeywfxatr2ia.png
In addition to the latitude and longitude, I've tagged it
"mashablecity" to indictate that it's part of our project, "geotagged"
to indicate that it has geotags, and added some descriptive tags that
we can use later, for instance to include this info in a map mashup of
just the cafes in the area. Additionally, I've put the literal street
address into the bookmark's description field, which could also come
in very handy later.
If we follow this convention, then we'll be able to retrieve an
aggregate list of all our geotagged Mashable City URLs at
http://del.icio.us/tag/mashablecity+geotagged/. There's already a
tool that can render this URL as a map at
http://worldkit.org/delicious/;
our map would be
http://worldkit.org/delicious/tag/mashablecity+geotagged/.
(It's not very granular, but it's a free head start. We'll be hacking
up a better map soon.)
I'm planning to hack together a quick mashup to assist with the
geocoding and tagging workflows, but in the meanwhile, I don't think
the manual process is so bad.
So, that's where we are right now. I'm really excited that we're
starting to actually build a useful corpus of info, and it's only
going to get more exciting when we can start the phase of extracting
the meaning from each of these URLs. In the meantime: Any feedback?
Anything we forgot?