Wellesley College researchers in Massachusetts have launched an interesting project dubbed Twitter Trails that attempts to show, via Twitter, how true and false stories propagate differently on the social network.
This one hits pretty close to those of us in the journalism business, where attempting to verify whether what we see on Twitter and other social networks is for real. And in fact, the researchers say on their blog that the tool is initially geared toward helping amateur and professional journalists to investigate recent and breaking stories.
ALSO: 13 of today's coolest network research projects
Examples on the blog include a story of a young girl who accidentally shot a gun instructor with an Uzi -- sadly, a real story, despite efforts by conspiracy theorists to make it appear untrue. Using the tool, researchers were able to show how accounts by verified Twitter sources propagated much more strongly than those of conspiracy theorists of dubious reputation.
Twitter Trails project The propagation graph, showing the tweets that were written when the story first broke on Twitter, all reporting that the shooting occurred. Various of the top tweets are labelled, coming from local, national and international news media accounts, and prominent journalists and bloggers. Also labelled is a popular tweet coming from a non-verified account commenting on the story.
Among the false stories examined were Netflix going bankrupt and a revelation that you can charge your iPhone in a microwave oven (DON'T!).
The National Science Foundation-funded Social Informatics Lab project (directed by Takis Metaxas and Eni Mustafaraj, and managed by Samantha Finn) relies on algorithms and visualizations to track Twitter trails. It also can track key words and phrases to show how various themes of a story live on after the initial news breaks.
You can dive into the gory details of their research paper here (TRAILS: A System for Monitoring the Propagation of Rumors On Twitter). These researchers are all over what makes social media tick, including how it can be used to predict election results.