This is more abstract/less worked out, but it might be cool to have a logistic regression function that can be used to relate one variable to another, in the data, to measure correlations; this is vague, and need to consider more: but, e.g., tracking correlations between creators/groups of creators and various keywords or topics from a topic model, etc.