Hi, Simon, On Wed, Oct 22, 2014 at 08:57:28AM +0100, Simon Nuttall wrote: > What options are available to group results that are effectively duplicates? > > For instance this search: > > > http://nominatim.openstreetmap.org/search.php?q=station+road&viewbox=0.13%2C52.2%2C0.14%2C52.19 > > produces as results 2-5 various small elements of the same Station > Road in Cambridge.
Nominatim doesn't group these road pieces internally as it does with place nodes and boundaries. It simply looks at the result list and filters exact duplicates. Those road elements don't count as duplicates under that schema because they have different postcodes. > The Nominatim FAQ at point 9 mentions: 'merge place nodes and boundary > relations to avoid duplication'. > > I'm wondering whether to do our own post-processing of the result to > group those cases together, whether others have tried something, or > whether there are other options? At the moment postprocessing is the only option. However, there have been quite a few complaints about this exact issue, so eventually this something that should be tackled in Nominatim, preferably by marking these street parts as true duplicates. > One other point: the above search viewbox is highly tightly wrapped > around the station road in Cambridge, yet the first result is a > polygon in Chingford about 80km away! That's because landuse has a higher rank than a road and higher rank always wins against viewbox bias. You could file a bug report about that specific case although at the moment I'm not sure that this can be solved. Sarah _______________________________________________ Geocoding mailing list [email protected] https://lists.openstreetmap.org/listinfo/geocoding
