Given a set of strings.

• Split each string into a list of substrings of at least three character each
• From the set of all possible decompositions, select the one with the lowest score

Scoring a decomposition

• Form the set of all the substrings (Keep only one of each duplicated substrings)
• Sum the lengths of all the substrings in that set.
• The score is that sum plus twice the number of substrings in the set.

I suspect that execution time of any solution would be exponential in the total number of characters.