Any best practices on including provenance information in APIs?

[MendeleyStack]

We are working a rather large data model across a rather large organisation. We have about 30 entities that form part of our overall entity graph. Examples such as Person, Organisation, Article, Patent, Conference, Journal etc. 

We are working on trying to improve how these data models are accessed but because we are a large organisation with entities spread in all manner of ways then the same data could be sourced from several sources. 

So my question is how do you design APIs where you have to provide provenance at field level ? 

For example we have an Article entity (think of a research paper) and that article could be sourced from 1 or more of our data sources because it appeared in more than one of our product suites. You would want to know where that underlying data comes from. 

I would love to hear to hear if anybody else has run into something like this?