[okfn-help] Geographical and temporal information in Open Correspondence

Thu Oct 28 09:21:56 BST 2010

Iain, very glad to hear of your progress, have a few points of reference

On 27/10/2010 21:40, print.crimes wrote:
> One of the things that I wanted to look at in Open Correspondence is the
> spatial and temporal information that exists in the letters.

I run a place-name text mining web service at the EDINA data centre, and 
would be more than happy to run some of your text and metadata through 
it and do a little analysis of the results. 
http://unlock.edina.ac.uk/text.html

The backend is developed by the Language Technology Group at the 
University of Edinburgh - they do some temporal/event text mining as 
well (for example, resolving "last Tuesday" to a specific date relative 
to the rest of the text, and connecting it to clusters of actions).
I forwarded your message to LTG folk, maybe it'll spark something.

> I had thought of trying to use Google maps to show the data for Collins
> but I'm sure that there is a better way of doing it.

OpenLayers is likely the best option for doing this with open source / 
data. http://openlayers.org/ - worth also looking at the new-ish 
http://polymaps.org/ that Stamen released with SimpleGeo for doing more 
complex visualisations with SVG overlays.

> The RDF endpoint does not expose the geographical data yet as I'm not
> sure of the best / most useful ontology to use though browsing there is
> some W3C work and opengeospatial has various vocabularies / ontologies

There's a GeoSPARQL standard going through the OGC right now. The effort 
is Oracle-led and so corresponds to their splinter of GeoSPARQL.
I'm not sure if OpenLink are involved - i should join the working group 
- but i've seen/heard good things about the geographic SPARQL query 
support in Virtuoso - it's being used for http://linkedgeodata.org - the 
RDF version of OpenStreetmap.

> It is not an area that I have much experience in yet (but it does seem
> that both MySQL and Postgresql support GIS to some degree though again
> better tools may exist)

PostGIS is the gold standard here, really. There's also a new wave of 
spatial extensions to NoSQL stores which, if you're into the fun of the 
learning experience, would be worth looking into.
GeoCouch for couchdb: 
http://vmx.cx/cgi-bin/blog/index.cgi/geocouch-the-future-is-now:2010-05-03:en,CouchDB,Python,Erlang,geo
MongoDB geospatial indexing: 
http://www.mongodb.org/display/DOCS/Geospatial+Indexing