LIterature on query generation from a labelled document term matrix

I have a labelled dataset of relevant and non-relevant documents for which I built a boolean document term matrix.

I am trying to develop an algorithm which given this input would create a text-based boolean search rule which identifies a subset of the data favouring first of all sensitivity and then specificity.

I'd like to know published literature on the topic. I made some initial search but couldn't find anything related. I'd be glad if you can point me to relevant papers/authors.