Great find, Rod, I remember doing 4,5-variable K-maps on paper, what a pain :-D
Combo, to answer your question, each group you select becomes a term of the final equation. You need to have all of the 1's (high outputs) circled within groups, and no more than that (or you will have redundancy - good in some cases, but not for the basic and stable PLC logic). In a 4-variable K-map (as in the app Rod pointed out), you can see what can be a group and what cannot. Basically, all the groups have to be rectangular or square (and they can also wrap around the edges as long as they satisfy that condition), you cannot have S- or L- shaped groups. Once you have all the 1's covered, you take each group, and see which of the 4 variables remain in a constant state for all the members of the group. The constants become your terms (ex. top left corner, group of 4, you have A and C which are always in a 0-state. So, that group becomes the term (NOT A) AND (NOT B). Each subsequent group (term) is OR'ed with the previous terms to create the final equation.
Hope that makes it a little bit clearer (and that's just the basics, you can get much more out of a K-map, such as debouncing filters, which you don't need in PLC programming anyway since your inputs are properly debounced by the input modules).
Finally, Simon, if you don't use "don't-cares" in a K-map, your final equation is logically (by Boolean Algebra rules) identical to what you start with, only much simpler. Only problem would is that it wouldn't normally show the logic of the rung as well.