What is the best way to implement reinforcement learning?

Question

Neural networks

I chose Q-learning in order to teach the bot to play tic-tac-toe (sports interest, although it doesn't really make sense)
I'm having difficulty.
1) How to store Q[s,a] ? Array, dictionary?
2) The previous state - the field, before the opponent's move?
3) The current state - the field, after the opponent's move?