Tweets reveal your age and gender
14 May 2013, 09:25
The Hague - Researchers at a Dutch university have
developed an online program that is able give the age and gender of users,
purely based on the content they post on the social network Twitter.
Based on data from almost 3 000 Twitter users who
"tweet" messages in 140 characters or less, researchers at the Twente
University near the eastern city of Enschede, have compiled lists of words and
sequences corresponding with different ages and specific genders.
Dong Nguyen, a doctoral student in computer sciences said
users simply had to enter their username into the online program, which then
calculated age and gender by comparing the last 200 tweets with the words and
phrases in its database.
"The distinction between men and women is actually
very stereotypical," said Nguyen, one of the participants in the project:
"Men talk about football and women about their nails."
"In terms of age, younger users talk about
themselves a lot more and use a lot of emoticons while older people use longer
words and sentences," she told AFP.
Emoticons are a number of icons such as a "smiley
face" to portray the writer's mood or facial expression in a tweet.
The program, which only analyses tweets in Dutch for now,
has a margin error of four years, which dips for younger users and increases
when users are older.
"We note that users use more uniform language from
about 35 years and older. There are larger differences between a users aged 15
and 20 then there are between users aged 45 and 55," Nguyen said.
Researchers at Twente in collaboration with the Meertens
Dutch language and culture institute, were looking at updating the program for
other languages and adapting for popular networking sites such as Facebook.