randomly take two different objects for one predicate

I have an RDF dataset like this:

<subject1> <some_predicate> "Value 1" .
<subject1> <some_predicate> "Value 2" .
<subject1> <some_predicate> "Value 3" .
<subject1> <some_predicate> "Value 4" .
<subject1> <some_predicate> "Value 5" .

<subject2> <some_predicate> "Value 6" .
<subject2> <some_predicate> "Value 7" .
<subject2> <some_predicate> "Value 8" .
<subject2> <some_predicate> "Value 9" .
<subject2> <some_predicate> "Value 10" .

Now, for each subject I want to have two random values of "some_predicate". They ought to be two different ones. So, the expected result would be something like:

--------------------------------------------
| subject | random_value_1 | random_value_2 |
============================================
| subject1 | "Value 2"     | "Value 5"      |
| subject2 | "Value 6"     | "Value 7"      |
--------------------------------------------

I found this question sparql: randomly select one connection for each node However, the problem there is just to get one value, I need two and different values.