Sweet, looks like something I could really dig into for unscatter.com I'm curious, are you proxying apis or crawling and keeping the information local to serve?
For social networks, we're proxying the APIs (essentially). For articles and news, we both aggregate organically discovered articles from streams, as well as scrape 250k+ RSS feeds and other news sources. So, articles are stored local to the server (which is why we have such a large collection of them), but sensitive social information is a pass-thru to the social network.
However, in both cases, we're doing lots of analysis on top of the data per our "normalization" philosophy.