The contents (individual values, names, etc.) of the data you upload into Lexio is not being used by Narrative Science. In order to train machine learning models that Lexio uses to intelligently guess about how source data columns map to the Lexio knowledge base, we may use column headers as part of those machine learning training models.
Note, Data is not shared across organizations. In other words these models are do not use data from other orgs in your stories, nor will data from your org be used for others.