Sean thinks the OGC-based GeoWeb isn't showing particularly stellar growth, press releases notwithstanding.
Google bears him out: http://www.google.com/search?q=inurl:request=getcapabilities
The number of results returned from this search isn't moving up all that fast.
Thursday, December 11, 2008
Subscribe to:
Post Comments (Atom)
10 comments:
I wrote once that I'd eat my keyboard if there were 10,000 domains serving WxS. 7000 looks pretty close, but Google shows only ~500 hits that are "unique". Here's a server with what appears to be many county-level services: http://www.google.com/search?hl=en&safe=off&q=inurl:request%3Dgetcapabilities&start=520&sa=N&filter=0. It alone appears to provide hundreds, if not more, of capabilities documents.
Or it could just be that there aren't a lot of links out there to the getcababilities links - none of the dozen or so WMS/WFS we run show up in the Google search. Though I did notice the googlebot rambling through our data sets this afternoon.
Isn't that a symptom of the overall disease, Ian?
why would I make links to my getcapabilities links?
So is the goal many servers or lots of data? I know of one WMS which has 800 layers (why is another story). That's valuable, even if through one endpoint.
I think content through WMS is out there. If folks wrapped their WMS through a KML gateway, you'd see this alot more apparent in search engine results.
@Ian Turton: Well that would be GeoWeb no? Everything is discoverable. OGC GeoWeb is poor where WxS might have deeper penetration. Who knows....
James - I guess but mostly I link to the client apps that display the data.
I'd about to download geoserver 1.7.2 which is suposed to plug directly into google indexing (http://blog.geoserver.org/2008/12/11/geoserver-google-maps-and-geo-search-a-request-for-data/)
Tom: there's data, yes, but discoverability is poor. And since WxS doesn't recommend use of URIs or links, network effects are small. I think it's practically impossible for this kind of architecture to "surge".
@Ian: Well that is the big problem with the GeoWeb. People who have servers need to decide if they want to expose their data to Google and make it discoverable.
I suppose companies like WeoGeo are attempting to make a marketplace around these discoverable data sources, but as of yet WeoGeo is not services.
GeoServer seems to be taking the lead (I'd throw ESRI in there with their RESTful API) in making their services available to be crawled. Now one just needs to decide if they want to do that.
We are pretty keen on letting the bots index our metadata and know about our services. In fact we encourage them. This allows more people to find the data and services we are trying to make available - through the means that they commonly use - search engines.
We did this by putting a simple dynamic sitemap onto our metadata application. http://ilmbwww.gov.bc.ca/geobc/services/metadata/news/#2008_04_03
Indexing our metadata content by robots has other advantages which go beyond getting knowledge of our data out there - it lets us take advantage of superior search algorithms and hardware than we can afford to - and use offerings like Google Custom Search.
We will eventually hook our metadata more tightly to our other services so that links from our metadata records will reach WMS / KML - however for now we are able to expand our 'web presence' by building routes for bots to climb into our services via manually crafted KML wrappers on our ever expanding WMS services posted to static webpages. See http://ilmbwww.gov.bc.ca/dm/wms
I am with Tom K. KML is indexed by the bots and increases our data and services visibility in search engines not to mention the little bonus that it makes it that much easier for the masses to use our WMS - which was of course the first objective.
Post a Comment