Can you learn about a person’s well-being through his or her tweets? Penn researchers have found a way.
The World Well-Being Project has collected billions of tweets and Facebook posts to study their correlation with personality, regional health and happiness — a project largely inspired by Google’s findings on the correlation between search queries on the flu data from Center for Disease Control.
One of the big components of the project is finding the relationship between personality and language in social media.
“We collected more than 100,000 people’s results of their Big Five personality tests” — openness, conscientiousness, extraversion, agreeableness and neuroticism — “and their Facebook posts,” said Lyle Ungar, a professor of computer and information sciences who is on the team. “We look at the co-occurrence of words and multi-word expressions, then we correlate them. How often does “sick of” relate to introverts, extraverts, well-being or cargo disease?”
One surprise Ungar had from the study was a strong correlation between being well-adjusted and participating in sports or outdoor activities.
“We don’t know whether it’s causal or coincidental, but it’s interesting to think about,” he said.
With a team of around 15 faculty, students and staff from various disciplines spanning from computer science to psychology, the project processes large amounts of data from social media and relies heavily on fast processing and efficient algorithms.
“There’s a very interesting mix of computer science and statistics that needs fairly fast processors and clusters because we have billions of tweets,” Ungar said. “It requires both some knowledge of statistics, psychology and just being able to write efficient code, which is why we have such a big team of people.”
The team is planning its next step to compare the relationship between wording and personality internationally, including comparison between the United States and the United Kingdom, and later even between the United States and China.
“We have undergrads or masters looking at questions: What can we tell from tweets? Can we distinguish people of the same age and sex? Is the definition of introvert and extravert different in the United States and somewhere else?”
“It is not completely obvious in the long run whether we have the same determinants of well-beings across countries,” Ungar said. “Being in a good relationship is universal, but other things are different. For example, religious people are happier than non-religious people in the United States. [This is] less true in Europe. My hypothesis is that in the United States, religious ties are deeply related to social ties. This sense of community increases happiness.”
The project has seen strong interest from industry, attracting attention from IBM and several startups.
“Companies are really reluctant in giving out their data, but they are very interested,” Ungar said. “A lot of companies want to monitor employee satisfaction because happier employees are more productive, especially in jobs [that require] creativity and ideas.”Comments powered by Disqus
Please note All comments are eligible for publication in The Daily Pennsylvanian.