2 years ago
TechAtHeart
in English · 2,068 Views
likes 4clips 1comments 2
Twitter Data Mining Reveals America's Religious Fault Lines
A team at Wright State University in Dayton, Ohio have analyzed more than 250,000 Twitter users in the U.S. who affiliate themselves with the following religions: Christianity, Hinduism, Islam, Buddhism, and Judaism. Here's some interesting findings in their study: - they "collected all of the words that appear in more than 100 tweets in each group and ordered them by frequency of use" ex: Christians use Jesus, Christ, gospel, pray, etc - "members of one religious group are much more likely to follow members of the same group than members of a different group." To be exact, following someone of the same religion is 646 times as likely than someone who isn't - "It is also possible to work out the most popular accounts followed by members of each group, and this reveals some interesting fault lines. For example, the most popular account followed by Buddhists is @Dalaillama , while the most popular among atheists is @RichardDawkins. All the religious groups are likely to follow @BarackObama, reflecting their American heritage." It's a pretty interesting study...though not entirely surprising, in my opinion. Some studies they want to do in the future include trying to analyze twitter data from areas of religious conflict. These data sets are probably huge, I'm glad that data mining is helping researchers understand our world through social media.
TechAtHeart clipped in 1 collections
2 comments
Not surprising to me, either, but I wonder how large of a sample the data came from? 250k doesn't seem like a lot considering the size of twitter-verse. Was that mentioned?
@onesmile Yes, it seems like a small data set size with the amount of twitter users online. They don't mention it, but my guess is that they created this sample from a larger data set...most likely clustering methods. You clean up your data by doing this and remove any outliers before running any sort of analysis on it.