How to handle data (do analytics) of various formats (unstructured data) in a unified manner for decision making (DM)

Question

Public sector departments are facing problems analysing unstructured data for DM. How handle data of various formats (unstructured data) e.g. text like sms, notes, emails, blogs images, etc using autoML with Rand Forest, ensemble & DNN for DM.

Not much but I could not figure out clearly the methodology where I pull together data from various sources and introduce it to a single model that is able to suggest the most appropriate machine learning (ML) model to handle the specific data for DM. In this case, I opt to use auto-sklearn together DNN, random Forest and ensemble. Output to automatically get best algorithm for particular dataset for DM