Geo-tags & Weblogds

At my last job we created technologdy for searching web pages geographically.  The way that it worked was that while a web page was being spidered and indexed the page content was also searched for items that contain geographical significance (zip code, phone number, etc..) which was then converted into an indexable form of Latitude, Longitude and indexed with the other keywords in the page.  Then, using a geo-enabled web search engine (Northern Lights powers one on Mapblast), you can find web pages that contain text AND are within some distance from your search location.  A prime example is searching for Bed & Breakfast places where you plan to spend a ski weekend.

I've since seen some mention of tagging weblogds with geo.tags and was even starting to add the meta tags to this weblogd.  On further reflection I decided it wasn't a good idea.  A geo.tag would be better utilized on my homepage or an "about me" page, since it is tagging my location and the location of my business.  But tagging weblogd pages with a meta tag means that everything I post, local news or not, is tagged under the umbrella of Lebanon, NH.  Not good.  Or, as the draft says:

The geographic position given is associated with the resource described by the HTML document, not with the physical location of the document [2], or the location of the company responsible for publishing or hosting the document.

If each entry in a weblogd lived on it's own page, then that would be a great place to add a geographic tag, when the logd entry has some geographic significance.   If a page could be tagged with multiple geographic tags, one for each subject, then that might help, but from my reading of the document it is a single location per page vocabulary.

Still, geographically tagging pages, when appropriate, is a good start and there are sites like geotags.com which are promoting it and providing tools for wider scale adoption.  Perfect for tagging your B&B or Snowmobile club page.

A few years ago I wrote a little program which regularly scraped (web speak for programmatically reading a web page and extracting content) a world news site.  Each news story had a location byline like, Athens, GA, which I would geocode (convert to lat/lon coordinates) and then build a world news map page with.   It was a fun exercise and seeing a world map dotted with icons for each news story is kind of interesting.  Unfortunately I only found one news source which consistently tagged their headlines and even then most of the news out of the US was tagged with the news bureau location (i.e. Washington, NYC).

I think the world of weblogds makes for a ripe means of taking this much further. Since many weblogds are also published and distributed in XML there may be the opportunity to add story specific location tags using some sort of geographic markup, like GML. Computerworld has an article on GML and more indepth information can be found at the opengis site. There are a few other efforts I uncovered in a quick google search this morning, including Dublin Core Metadata Initiative which has a Coverage tag in it's DCMES.

All of this is probably much more involved than the average weblogdger wants to consider when posting a link. Heck, most folk's threshold of frustration is more than exceeded when trying to turn a location into a latitude/longitude: Why doesn't it know where my street is...! To succeed a geographic tagging system will need to be turned into an easy UI, not a simple task at all, at least from my experience maintaining Mapblast for five years.  So far the simplest (from the publisher's point of view) is probably the location byline that you see in newspapers:

    Paris: Investigators still have no clue into the....

It's probably easy for a weblogdger to put this at the front of their post or even within a hidden tag if the posting interface (blogdger, radio, etc..) provides a means to do so. Asking them to convert it into a valid latitude/longitude probably exceeds the annoyance threshold for ninety five percent of the weblogdgers. Without a standardized methodologdy this puts more burden on the search/indexing engine: did they mean Paris France or Paris Texas, was this a post by Paris Smith or part of the script from Paris Trout?  That validates the argument for a story-based geo-tag, since the search engines which deal in the XML (RSS) of weblogds don't have to reverse engineer the textual content, but simply pick out an embedded tag.  The RSS for my weblogd looks like this: JerZone RSS.

In looking at my page's xml I can see a story posted about Lincoln, ME which could instead look like:

- <item>
<geo.placename>Lincoln, ME</geo.placename>
<description><P>Ah, finally, a way to
<A href="http://www.sciencedaily.com/releases/2002/01/020123080138.htm">deodorize paper mills</A>.
  I remember seeing a bumper sticker while travelling through Lincoln, ME (a stinky paper mill town)
which said:</P> <P align=center><STRONG><FONT color=green>Kiss Me Where It Stinks: Lincoln, ME</FONT>
</STRONG></P>
</description>
</item>

It's not the ultimate solution, but it is pretty easy to understand and might start things down the road towards a more comprehensive system of geographical markup.

I'm a map freak and can easily imagine creating a couple of dynamic maps tied into the weblogd community. 

It would be interesting to see a map that shows icons on a world map for all of the recently updated weblogds.  In this case you would need the geographical coordinates (approximate coordinates, zip/city, is fine) for each weblogdger.  This would be included with the meta information, kind of like the Channel headers <b>title, link, copyright</b> in the RSS feed.  You can imagine this map having icons scattered all over it, but the predominate bunch would travel, wave-like, in syncronization with the sun.  Using this kind of map there's a good chance you'll find other weblogdgers in your area that you might not have discovered by other means.

The other interesting map would show the locations of entries posted in weblogds.   Most entries probably won't have a geographical tag, but when you post a geographically relevant item (i.e. Lincoln ME, Afganistan, Washington DC) it would show up on this world map.  It might not even be news, but a report on someone's camping trip, review of a new restaurant, or even a garage sale you're having.  It's difficult, if not impossible, for me to search all of the weblogds for information which might be locally interesting, but if they are geographicall tagged then I could do a proximity search or view recent posts visually on a local map.

 


Other Stories and Photos - The Jer Zone



© Copyright 2002 Jerry Halstead.
Last update: 4/27/02; 9:01:13 AM.