awk 'BEGIN {FS = OFS = ";"}
FNR == NR {
        # Save the 1st and 2nd fields from each input line...
        l[NR] = $1      # Save Leftmost field for output.
        k[NR] = $2 OFS $3 # Save middle output field and Key for matching.
        # Split the 1st field on percent characters.
        n = split($1, a, /%/)
        if(!(kc[k[NR]]++)) {
                # This is the first time we have seen this key.
                # Set the Count of fields for this key and set the min & Max
                # values for each subfield to the values on this line.
                c[k[NR]] = n
                for(i = 1; i <= n; i++) {
                        # Adding 0 makes the comparisons work even if some of
                        # the values are negative, and allows us to just print
                        # strings at the end without needing to use sprintf %d
                        m[k[NR], i] = M[k[NR], i] = a[i] + 0
                }
        } else {# We have seen this key before, update the min and max values
                # for each currently saved subfield.
                for(i = 1; i <= c[k[NR]]; i++) {
                        # formats to convert empty fields to zeros.
                        if(a[i] + 0 < m[k[NR], i]) m[k[NR], i] = a[i] + 0
                        if(a[i] + 0 > M[k[NR], i]) M[k[NR], i] = a[i] + 0
                }
        }
        if(n > c[k[NR]]) {
                # We have more subfields than we have seen before for this key.
                # Set min and Max values for the additional subfields.
                for(i = c[k[NR]] + 1; i <= n; i++)
                        m[k[NR], i] = M[k[NR], i] = a[i] + 0
                c[k[NR]] = n
        }
}
END {   # Set the min/Max output field for each key we have seen...
        for(i in kc) 
                for(j = 1; j <= c[i]; j++) {
                        mM[i] = (j > 1 ? mM[i] "%" : "") m[i, j]
                        if(m[i, j] < M[i, j]) mM[i] = mM[i] "-" M[i, j]
                }
        # Print each input line with the added min/Max values for each subfield.
        for(i = 1; i <= NR; i++) print l[i], k[i], mM[k[i]]
}' file

1%2%3%4%;AA;1;1-5%2-6%3-7%4-8%0-9
5%6%7%8%9;AA;1;1-5%2-6%3-7%4-8%0-9
1%2%3%4%5%6;BB;2;1-7%2-8%3-9%4-10%5-11%6-12
7%8%9%10%11%12;BB;2;1-7%2-8%3-9%4-10%5-11%6-12
13%14%15%16%17%18;BB;3;13%14%15%16%17%18