I don't think online learning is possible for ALS in Spark. That means you can't update the model while getting the data in real time. However, you can use the model to get the predictions.
Also, refer to:
How to update Spark MatrixFactorizationModel for ALS