Normalizing Textual Data with Python