Tom Kyte has a good mantra for this, that goes something like this;
"not all full table scans are bad, not all index scans are good"

Why do you want to reduce the full scans? Is anyone complaining about
performance? If so, start with them and see what process they are
undertaking. Probably use an extended trace on their session and
understand what is causing the bottleneck.

If you are just trying to be proactive, perhaps start with Statspack
and examine SQLs that are performing a lot of buffer accesses but are
only returning a small number of rows. These SQLs may perform FULL
table scans, or they may not. Some of the worst performance problems
I've seen have been from forced index scans.