Re: Help needed for ada package

[tmoran@acm.org]

>Potentially, it is possible an individual is associated with everyone
>else hence there being one group of one million or 999999 pairs of groups.
  "Group" is misleading here.  Your "procedure Associate" leads to a
symmetric boolean N x N matrix where M(I,J) is True iff person I and
person J have ever been associated.  Finding the worst citizen in the same
"group" as citizen X means finding everyone X knows, then for each of X's
associates, finding all their associates, for each of those finding all
their associates, etc etc, and noting the worst person you come across.
If this is a brand new police database, then there are probably very few
known associations, so that's not too terrible in either storage or time.
  If you've been monitoring all associations between any two people over
the last many years, everybody is "associated" with several hundred
others, each of whom is associated with several hundred, and so forth.
Evemtually you'll find that, as in "six degrees of Kevin Bacon",
everyone except for a few cloistered nuns is associated with Kevin Bacon,
and thus with each other, so there are only two groups - those citizens
"associated" with Kevin Bacon and those in the nunnery.  Assuming there's
someone more dangerous than the nuns, Next_Member will simply be an
enumeration of all non-nuns.  Is that useful?