Re: City name 'spell checker'

Your problem is that city names don't conform to anyone's spelling rules.

If your only problem with zip codes is that some zip codes represent multiple place names you could use a two table lookup scheme. The first table would consist of three columns; Postal Code, Place Count and, if Place Count = 1, Place Name. The second table would consist of three columns; Postal Code, Sequence, Place Name.

The user would enter a postal code and either receive back the place name or a list box containing the appropriate place names.

hth
jerry gitomer

Keith Morgan wrote in message <82947k0hmo_at_enews4.newsguy.com>...
|Thanks for your reply. The 'city names' may not be mappable to
zip codes
|because they don't necessarily conform to post office standards
and I've
|found soundex lacking in accuracy. I am really looking for
something closer
|to a spell checker.
|
|Aubrey Soper wrote in message <3847FFFD.F41916A2_at_bigfoot.com>...
|>hello
|>
|>how about using a zipcode database, either to get the city
|>from the zipcode, or to use soundex on the city names
|>[limiting queries by state]?
|>
|>regards
|>aubrey
|>
|>Keith Morgan wrote:
|>
|>> I want to provide a city name spelling checking and
suggestion facility
|for
|>> a data entry application. I don't need the advanced
linguistic analysis
|of
|>> natural language spell checkers, but I would like to provide
a suggestion
|>> list of possible corrections.
|>>
|>> I am leaning toward producing a suggestion list based on edit
distance
|(AKA
|>> Levenshtein Metric). The edit distance is easy enough to
compute but I
|>> haven't figured out how to efficiently access likely
candidates to apply
|the
|>> criterion without a full table scan.
|>>
|>> I am open to any suggestions, references, or other ideas.
Thanks for your
|>> help.
|>>
|>> Keith Morgan
|>
|
|
Received on Fri Dec 03 1999 - 23:28:33 CST