On Sun, Feb 8, 2015 at 8:23 AM, Greg Troxel <[email protected]> wrote: > > 40m seems too close for fuzzy matching, or rather I think you need two > thresholds. I would think that you'd want something like 250m to flag > it as sort-of-matching for human review.
250m is slightly too large, as several campus installations are that close. The goal is that the OSM mapper is free to move stations up to (threshold) meters, and the conflation step will just assume the OSM mapper is correct. The idea was that over 40m would trigger human review to see if the upstream data source needed correction. In general, since we're talking about a feature too small for air photo review, the armchair mapper has to trust the local mapper. Perhaps you could run the conflation script and post the osc for > review. There are so few stations in OSM now, there's no interesting result from the conflation: the main overlap are stations I added myself to test conflation. The conflation is really about maintaining this data set in the future: keeping a commercial dataset in synchronization with an open source dataset.
_______________________________________________ Imports mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports
