Yes, I’ve been working on this for a good while. Text normalization is basically the massaging of data in various ways. I have found rule-based regex engines most practical for this. I have been trying to convert fragmentary tweets into proper sentences, in order to input into a dialog system knowledgebase for question answering, and thereby create a dynamic knowledgebase crowd-sourced from tweets. In terms of Twitter bots, Twitter is very sensitive about people ripping off other people’s tweets; therefore, it’s necessary to alter the content in this way to avoid “copyright” issues.