I recently began working on a (somewhat) related NLP project looking at the shift in sentiment in Trump-focused articles published pre- and post- election. The motivation for the project is the observation that many of the left-leaning news media outlets - who consistently lambasted Trump in the lead-up to the US election - have begun to dial back their criticism post election, presumably in an attempt to (re)build bridges and ensure their continued relevance in the ensuing Trump milieu. It seems that the results of such an analysis have the potential to be a concrete and relatively simple example of the deviation from any stable media narrative, and perhaps a nice opportunity to spread this message to a slightly wider audience.

The project is nascent, but it should be straightforward to implement (I have already begun to amass articles from several major news sources). While this may or may not be relevant to your stated goal, I'd be happy to share more info if you're interested!