List comprehensions can do this easily, esp when you leverage that True == 1 and False == 0:

n, m = raw_input().split()

sc_ar = raw_input().split()

A = set(raw_input().split())
B = set(raw_input().split())
print sum([(i in A) - (i in B) for i in sc_ar])

Instead of using multisets, its enough to use sets on A and B since set lookup is O(1) in python (Awesome, right?)

Here is Python 3 solution from my HackerrankPractice repository:

_ = input()
array = input().split()
like = set(input().split())
dislike = set(input().split())
print(sum((i in like) - (i in dislike) for i in array))

Feel free to ask if you have any questions :)

The problem is quickly resolved by using list comprehensions. This is my solution in Python 3:

n, m = input().split()
arr = [int(x) for x in input().split()]
A = set([int(y) for y in input().split()])
B = set([int(z) for z in input().split()])
count = [0 + 1 if x in A else 0-1 if x in B else 0 + 0 for x in arr]
print(sum(count))

I think the best way to do this is by using multiset (called as Counter in Python, this can be found in the collections module). For more info read this link. First store the input data in the form of list. Then use this list (an iterable) to generate the counter object. This object will have a key-value pair.

Further, you need to iterate through the two sets and find the value in the counter object and increase / decrease happiness.