Need help idea/suggestions to implement an architecture multi data sourced environment for different proportion of time

Hi considering the following data flow structure:

I need ideas/suggestions to implement a smart architecture that attend to load data from different sources with different structures and different formats (xml, json, plain text) that contains the same business information (sometimes one of the sources come with a wrong data and it must be replaced with that piece of data from the other source ), decode, normalize to a common business fact, considering an clouded code infrastructure and an API that works with the generated facts.

I was thinking on something that converts any format to a kind of hash (That should be stored as raw data) and use a kind of strategy implementation to map each field data to an business entity and for each strategy, supply a chain of responsibility objects that are responsible to sanitise the data before store in the entity. I'm trying to find some similar situations and examples but every idea is fragmented.

Another thing is the time control, i didn't found some good solution to manage the time of fact (i guess), i was thinking on date time field based strategy but i need to work with something more seamless ( or uncouples ) because we work with different providers and rules can change easily.

Can you guys help me with some advises, ideas?

Any suggestions, examples i'll be happy to read and discuss with more details.