Matrix shorting into fundamental sets

Hello,

My question is how to short a matrix of percentages into fundamental sets that have sub-sets that represent the percentage contribution into that set. Example:

This is my input matrix:

   obj1 obj2 obj3 obj4

obj1 | 1.0 , 0.0 , 0.0 , 0.0

obj2 | 0.0 , 1.0 , 0.4 , 0.0

obj3 | 0.0 , 0.4 , 1.0 , 0.6

obj4 | 0.0 , 0.0 , 0.6 , 1.0

I want to compare the objs:

---------------------------------------

obj1: obj1 shows no likeness to the other objs.

obj2: is 40% like obj3.

obj3: is 40% like obj2 and 60% like obj4.

obj4: is 60% like obj3.

---------------------------------------

two sets come out of this:

1) obj1 is all by its lonesome : [obj1]

2) since obj2 is like obj3, obj3 is like obj2 and obj4, and obj4 is like obj3: [obj1 obj2 obj3]

---------------------------------------

But all obj in set 2 are not equally alike, so I want to subdivide this set into subsets that describe how alike they are. Since there are only two percentages this comparison is simple. (but in actuality the percentages will range from a define cutoff to 1 [example:0.80 - 1 | this shows all objs at are at least 80% alike in the initial matrix] I would like to be able to input n number of subsets).

number of subsets (n) = 2;

veryAlike (n_1) = [obj3 obj4];

Alike (n_2) = [obj2 obj3];

final sets:

{[obj1] , [[obj2 obj3] [obj3 obj4]]}

So I can look at the output and see that set1 = obj1 and set2 = obj2,obj3,obj4, and if I would like to see how much objects in each set show good comparison I can easily pull that information out also.

Thank you