RNFdev Southern Ontario FSA Border Streets

[Dan Putler]

Hi Dave,

I don't know if he has joint the list, but he sent me email on this  
issue. To that end, could you send me the original address file for  
Ottawa so he has something to test with? My guess is as to what will  
work best is if each road is placed in a single data value. For example:

LANDSDOWNE RD S

Given that PAGC's I/O is based on Frank's Shapelib, things will work  
best if this single field database is saved in a dbf file. Walter can  
then produce a standardized file that conforms to the StatsCan RNF  
structure. Road names are another issue. However, we can use a set of  
unique road names taken from the RNF for an area and then match the  
Canada Post road names to them using PAGC's existing matching routines.

Once I get the Canada Post Ottawa file from you, I can quickly  
generate the StatsCan RNF road names and them along to Walter.

Dan

On 25-Sep-06, at 5:48 AM, Sampson, David wrote:

> Hey Folks,
>
> I am taking a look at southern Ontario, specifically St Catherines  
> and GTA.
>
> I tried to come up with a systematic search and replace for St  
> Catherines and only had a few instances that I needed to check the  
> PDF for varification. Ther are not many names on that map.
>
> The GTA though is a whole new ball game (20 pages at 12 pt font  
> after removing all FSA codes)….  Within the first page I would say  
> close to a %60-70 check back ratio, where I needed to check the  
> original PDF and the names did not appear in a predicatble manner.
>
> For instance prefix or suffix would apear 2 to three lines previous  
> or after the name of the road. And also some names were split due  
> to length. It is less teadious when names and prefix/suffix occur  
> on sequential lines. One process I used was to replace any manual  
> line break with nothing, making the whole document one continuous  
> text string. And then replacing key SUFFIX's with a coma (,) and a  
> line break. The comman was a reminder than the name has been  
> process, and the line break makes it an independent record.
>
> I would use common ones like RD, ST, CIR, AVE etc and for smaller  
> areas this worked quite well for a process, but had no means of  
> verifying if the names are what they are supposed to be.
>
> I explored the concept of using th RNF DBF as an Excell Verticle  
> lookup table, but Excell is limited after that. A Vertcile lookup  
> table can be used to check a given value (eg suggested road name  
> text string) against a standard list. This is a dirty approach but  
> I can keep exploring it.
>
> Does Walter have any ideas as to how we can process the raw names  
> and try to standardize them?
>
> Although it IS possible to manualy match street names for the GTA  
> this would be nothing less than a Nobel task.
>
> Cheers
>