word score

[jwelch]

Is there a way that I can create a query that will return a word approximate score such as: (refering to a table with first_name (text), last_name (text), ...<br>

<br>

A user could enter

 

[DougP]

Huh

 

[jwelch]

Sorry, for some reason the last post got truncated.<br>
<br>
<br>
Using SQL with a ODBC source which is a Access2000 database...<br>
<br>
Is there a way that I can create a query that will return a word approximate score such as: (refering to a table with first_name (text), last_name (text), ...<br>
<br>
A user could enter &quot;Colinsworth&quot; as a parameter and the query would return all records that had an exact match of Colinsworth (which would be 0 records in this case with a score of 100%). <br>
<br>
Also, it would return the records of &quot;Collinsworth&quot; at a very high score and records of &quot;Collins&quot; at a medium score.<br>
<br>
I understand how to set up SQL queries using parameters, I just don't know how to score word approximatly.<br>
<br>
Thanks

 

[DougP]

What is the underlying result here.<br>
Sometimes in these forums its difficult to gather what you are really after.<br>
What purpose does it serve to know what percentage of someones name is in the whole group of last names?<br>
is this statisical anyalsys. <p> DougP<br><a href=mailto: dposton@universal1.com> dposton@universal1.com</a><br><a href= > </a><br>

 

[jwelch]

The user would enter a first and last name or just one of the two, and the query would return records with a score of how close the fname and lname matched. <br>
<br>
If there is an exact match of first name and last name then the score would be 100%, if only first name or only last name then 50%, if the first or last name is approximately (a possible typo or misspell) then it would be give extra points and a higher score.<br>
<br>
Example:<br>
user entered: Ed Collins<br>
<br>
If there was a Ed Collins then I know how to find the exact record.<br>
<br>
if there wasn't an Ed Collins, then it would score the following names:<br>
Edward Collins = 75%<br>
Edward Smith = 25%<br>
Ed Collinsworth = 75%<br>
<br>
etc, <br>
<br>
this is so that you can enter a piece of someone's name or misspell it and still get a few records are are about (about is the command for this in oracle and there is a similar command in sql server, i believe).<br>

 

[jwelch]

the underlying result would be a name search that would return the closest name if an exact name isn't found