[Talk-us] Remap-a-tron level 2 complete! Suggestions for level 3?

[Mike N]

On 9/28/2012 11:28 AM, Martijn van Exel wrote:
> That would be a pretty complex query, for the following reasons:
> * Similar naming may be obvious to a human observer, but can be really
> hard to detect algorithmically.
> * Similar geometry can be hard to detect because of split / combined
> ways so you would have to go segment by segment, or something.
>
> I'm happy to take suggestions for a query to make this work though, I
> agree it would be useful.

   The similar naming may be a difficult problem, but how about TIGER 
2012 ways with names, while the OSM way which includes this section have 
no name at all?

    An advanced case is to join TIGER ways having the same road type and 
name, then compare those to OSM.   I don't think this is necessary 
however the purposes of a rematron.   It will create a number of 
separate sections in an area to be called out as work units, which are 
generally mapped from Bing as a single way.  Perhaps nearby sections of 
the same name could be checked for and combined when a work unit is 
selected.

 >How many of those are there? They would only be candidates if they
 >actually have names. How would we know this?

   I also saw many driveways / tracks originally set as residential 
before redaction on the last Remapatron level.    It's frequently hard 
to classify these from even good aerials alone.   The Name/No Name test 
above will at least call out those having TIGER names and those without 
TIGER names are best left to locals.