On 23/09/2010 9:25 AM, Andreas wrote:
> Hi,
>
> how could I divide streetnames from housenumbers ?
You can't, reliably. There are papers written on the subject of
addressing, address analysis, addressing in databases, etc.
How would you handle the address:
Person's Name RD3 Clemo Rd Whangarei
? "RD3" is "Rural Delivery Area 3". The posties deliver within the area
by named recipient. This is a real address scheme.
The world is full of weird and wacky addressing. IMO, unless you're
willing to confine your schema to only handling addresses of a
particular area you know about, don't try to normalize address. Even
then, I wouldn't try to normalize addresses with text processing, I'd
ask the user to do it during form entry or not do it at all.
If you're trying to find duplicate addresses, matching addreses, etc,
then IMO you're better off using existing tools that do this with
free-form addresses using national phone databases, postcode databases, etc.
--
Craig Ringer
Tech-related writing at http://soapyfrogs.blogspot.com/
