i agree w/jZed's response to this, too.. For me, i basically had 3 reasons (some of the overlap w/jZed) for the DBI suggestion:

• i naturally thought DBI beacuse of how the OP worded the requirements -- it screamed RDBMS, and i strongly suspect the OP whad SQL in his mind when envisioning the requirements.
• The DBI solution could easily turn out very elegant, as opposed to a slew of (possibly tedious) loops/hashing up (though it may be a good excercise for OP)
• The scope/size of OP's problem was unclear -- it seemed like OP might have just given us a very small example, and could very well have a lot more data (and "tables"), which goes towards making the DBI solution more appealing. (and faster once it's set up, cause of indexes)