Efficient generation of random data subsets

[Tony Stephens]

I have a database table that contains a set of numbers (set X). I need to be
able to generate/export a random set of survey numbers (set C) from set A
with IDs
that aren't in set B. Where set A is a subset of numbers from set X that
match the filters for the survey, and set B contains the set of number IDs
for numbers that have already been exported for the survey.

The idea is to repeatedly generate a random sample of non-duplicate numbers
from set X filtering on the survey templates.

My database comprises of the following tables:

Table Numbers
ID INT PK
Number VARCHAR

Table Templates
ID INT PK
Template VARCHAR

Table Survey
ID INT PK
Name VARCHAR
Description VARCHAR
Started DATETIME

Table Filters
ID INT PK
SurveyID INT FK (Survey:ID)
TemplateID INT FK (Templates:ID)

Table NumberExportedS urveyJoin
ID INT PK
NumberID INT FK (Numbers:ID)
SurveyID INT FK (Survey:ID)
Exported DATETIME
This is the SQL query that I'm using to generate set C (the random set of
numbers) :

SELECT * FROM Numbers WHERE Number LIKE 'filter1%' OR Number LIKE 'filter2%'
AND Numbers.ID NOT IN (SELECT Numbers.ID FROM Numbers LEFT JOIN
NumberExportedS urveyJoin ON Numbers.ID = NumberExportedS urveyJoin.Numbe rID
WHERE NumberExportedS urveyJoin.Surve yID = S) ORDER BY RAND() LIMIT N;
The problem is that MySQL takes a long time to generate set C even for
moderate sizes of set A and set B. As the number of entries in table
NumberExportedS urveyJoin (set B) increases so does the time to generate the
next
random set of numbers. Is there a better/more efficient approach to
generating set C ?

Thanks,

Tony