The analysis of 75 million tweets from 189,000 Twitter users proved that like-minded people tend to use a common language and group into ‘tribes’. The use of certain words revealed a behavioral pattern able to map people according to their beliefs, interests, careers or ethnic groups.
The researchers used algorithms to monitor the exchanged messages and group the users into tribes. In their graphic representation (here) the words present are the most widely used in the community, while the circles represent the communities according to their size and activity.
For example, fans of Justin Bieber have a liking to end words with ‘ee’ (as in ‘pleasee’), animal welfare groups use words like ‘anipals’, ‘furever’ and ‘pawsome’, teachers tend to use long words while African Americans tend to shorten words (as in ‘chillin’). The list goes on with abbreviations and made-up words reflecting the backgrounds and preferences of various groups.
According to Dr. John Bryden, one of the study’s authors, “…by looking at the language someone uses, it is possible to predict which community they belong to, with up to 80 percent accuracy.” The team now says that can work out the tribe of an individual by analyzing their tweets. They actually believe their method can have commercial applications and have already applied for a patent.
Figures, data and full article: here