#mediaanalyhsis — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #mediaanalyhsis, aggregated by home.social.
-
I've got this to about 60% of posts classified (by submitted site). I can continue winnowing this down, though there's obviously diminishing returns.
I've also revised my analysis code so that anything that's not classified defaults to "UNCLASSIFIED", without having to explicitly code that in the sites file.
I'm thinking of how I might crossref / correlate the site-based findings with title-based analysis. I'm also thinking of looking at average comments / votes by classification, as well as looking at the ratio of comments to votes (HN uses this as a very rough "flamewar" heuristic, though on somewhat shaky grounds IMO).
My sense is that many of the less-frequently-posted sites will turn out to be blogs of some form. I'm thinking of how I might assess this without having to key all of them.
<stage_whisper> random sampling <\stage_whisper>
One issue issue for less-frequently-occuring sites is that it's easy to code those which match a pattern (twitter, blogspot, livejournal, medium, substack, etc.) than those which are idiosyncratic. Note that a lot of Medium blogs don't appear on Medium domains, as well.