Tracking Breaking News across Social Networks

Gothenburg.
The next speaker at AoIR 2010 is Luca Rossi, whose interest is in how information propagates through social network sites. This works with data from Friendfeed, which is somewhat similar to Twitter, but also allows people to add their own comments and likes directly to other’s posts (more similar to Facebook in this regard).

How can we define information propagation on this site, then? If a user posts some content on Friendfeed, then this message is visible to all of their followers – and if one of the followers comments or likes that message, it also becomes visible to the people following that follower, and so on. Luca’s team collected Friendfeed data for two weeks in September 2009 – some 10 million posts including 500,000 likes from 450,000 users.

From these, they selected Italian-language posts only (some 200,000), and calculated how often users posting in Italian also posted in other languages – Italians were very consistent at posting in their language, while Scandinavians, for example, posted far more often also in English or other languages.

The project also examined what happened during acute media events – for example, following the death of famous TV host Mike Bongiorno. His death caused a spike in messages on the day of his death as well as on the day of his funeral; the first entry related to his death generated some 130 comments, with nearly 600 comments overall.

Luca’s team combined information on post timestamps and follower networks to track how information about Bongiorno’s death travelled across the network – showing on the one hand some lengthy propagation chains, but also a few isolated entries which were caused not by information discovery through Friendfeed but triggered by reporting in other media (and indeed, many of these isolated entries were simply the imported feeds from major newspapers and other news sites. There are different activities at play here – explicit news sharing, with follow-up discussion and chat; over time, explicit news sharing turns into implicit news sharing and a public ritual of mourning here.

Breaking news is very important for tracking this. We don’t just use all of our network connections in the same way; during moments of breaking news we might utilise other connections in our network than we might do during general online interaction.