From reading this and this, I understand that a hidden service builds a virtual network to an introduction point and provides it with its public key.
What I'm wondering is, how does the introduction point then communicate with the hidden service using this information?
(as the second link says we do not want these nodes to know the hidden service's IP address)
The hidden service builds a circuit just like a normal tor client does, passing through 3 different nodes to connect to it's introduction point.
It is nowhere explicitly mentioned in the tor protocols how comes that an introduction point can still contact the hidden service if time has elapsed since the service has chosen them, but I think it is implied here that the hidden service keeps it's connections to introduction points open.
That means also that if the service goes down temporarily, when going back up their tor client will reconnect to the introduction points and if it doesn't there shall be no way and no need for the IP to find the service back.