New algorithm predicts which photos go viral on Facebook

At the upcoming International World Wide Web Conference, the Standford team, which has developed the algorithm, will elaborate on how the algorithm works

Published Date
07 - Apr - 2014
| Last Updated
07 - Apr - 2014
New algorithm predicts which photos go viral on Facebook

A team of computer researchers from Stanford have created an algorithm which can measure the chances of a photo going viral on Facebook based on how shares play out. The team will explain how the algorithm works in the upcoming World Wide Web Conference.

The scientists studied 150,000 Facebook photos, and found that 1 in 20 photos posted on the social network gets shared once only. And just 1 in 4,000 photos gets more than 500 shares.

Stanford researchers said that the clues to predicting which of the photos on Facebook go viral, lie in ‘cascades’. The term cascades is used to describe photos or videos being shared multiple times. The team was able to predict (with 80% accuracy) as to when a photo cascade could double in shares.

The scientists looked at different variables that might help them predict doubling events more accurately, including the speed and rate at which photos were shared, and the structure of sharing. They found that photos re-posted on multiple networks proved to create a stronger cascade.

“It wasn’t clear whether information cascades could be predicted because they happen so rarely,” said Jure Leskovec, assistant professor of computer science. “Slow, persistent cascades don’t really double in size,” he added.

Facebook had recently revealed that it reached 1 billion active monthly users milestone for mobile phones and its subsidiary, Instagram also reached 200 million users milestone. The numbers were revealed by Facebook CEO Mark Zuckerberg during an investor conference to discuss its latest acquisition of virtual reality firm Oculus Rift.

Source: Stanford