Provides normalizer for text. E.g. takes a text and checks for umlauts written as "ae", "oe", or "ue" and normalizes them if they really are umlauts depending on a frequency model.