The municipality of Nijmegen maintains an RSS Feeder for information it publishes. Now it wants to provide a spatial selection service to local content providers and citizens. The information behind the RSS Feeder is not yet geocoded, so the service must try to extract the geographic location from the content of the published information.
Nijmegen started by creating a simple list of named places such as neighbourhoods, parks and other well known locations. The service then scans the published text for names from this list and adds a geographic tag for each name found. This scheme worked quite well, but as the size of the list grew, it became slower while the number of newly found names in documents was increasing only very slowly. This is when Nijmegen decided to ask Geodan to have a look at the problem.
The most obvious way to improve the geotagging process was to increase the list of named locations and create indexes for search. Nijmegen has a complete list of all addresses available in their ‘BAG-database’. They also created a list of well known ‘points of interest’ from other sources. Geodan combined these sources into a hierarchical database where larger geographic entities are ‘parents’ of smaller entities. This way, the province of ‘Gelderland’ becomes the parent of the municipality of ‘Nijmegen’ which itself is parent of the stadsdeel of ‘Dukenburg’ etc. The database contains geographic entities from the country level down to the level of (well known) individual buildings.
Published documents are scanned for names in this hierarchical database and tagged with the corresponding geographic entities. Users can now select these documents using these tags. When users supply the name of the area of interest, this name is converted to a tag in the same hierarchical database. Now the database can perform an indexed query for documents that have tags equal to or inside the supplied tag.
Since the documents are now geotagged, the RssFeed could in principle be displayed on a map. This is possible by extending the RssFeed to GeoRss. Among others, Google Maps has the possibility of directly displaying a GeoRss Feed on the map. To see how this works, see example “d” below
Some results to be viewed online:
a. A Nijmegen RSS feed for press releases: http://www2.nijmegen.nl/RSS/persberichten
b. The same feed filtered for info about stadsdeel ’Dukenburg’: http://geotagger.geodan.nl/FilterRssFeed/FilterRssFeed?toponiemen=dukenburg&url=http://www2.nijmegen.nl/RSS/persberichten
c. The same feed filtered for point of interest ‘Het Valkhof’: http://geotagger.geodan.nl/FilterRssFeed/FilterRssFeed?toponiemen=het+valkhof&url=http://www2.nijmegen.nl/RSS/persberichten
d. The result can be displayed on a map as GeoRss.
Visit http://maps.google.nl and in the “Search in Maps” field enter the following URL for Dukenburg articles: