I have a set of documents like a list of memory register information, a list of pinout information and so on, for different types of equipments that are installed in different facilities. Each facility may have 5 equipments of different types. It sums up 60 facilities in total and so far I have created one set of these documents for 20 facilities. The number of variation of these equipments may increase as I move on, but I can say I have a good sample of what might comes ahead. Is there a way to create a machine learning model which I would throw in these already made documents (consider these documents of the same format) to train the model and to improve this model as it goes, so all I have to do from this point on, besides creating new documents for recently discovered documents and retraining the model, is telling the model what type of equipment I want to generate the set of documents and it comes up with what documents I should pick?
What type of model should I go for? Is there any paper or someone who has done something similar?