Scientists develop computer algorithm to predict viral sharing of photos on Facebook

Scientists have developed a computer algorithm that predicts whether a photo would go viral from how fast it was shared on Facebook.

According to Stanford researchers, the clues to predicting which of the many millions of photos on Facebook would spring from obscurity and go viral lay in 'cascades', the term used to describe photos or videos being shared multiple times.

PTI quoted Jure Leskovec, assistant professor of computer science as saying it was not clear whether information cascades could be predicted because they happened so rarely.

According to data provided by Facebook scientists in a recent collaboration with university scientists, only 1 in 20 photos posted on the social network got shared even once, while only 1 in 4,000 got over 500 shares – a lot but hardly an epidemic.

In a paper to be presented at the International World Wide Web Conference in Seoul, Korea, the researchers would share how they accurately predicted, 8 out of 10 times, when a photo cascade would double in shares; that is, if a photo got 10 shares, would it get 20? If it got 500, would it reach 1,000, and so on?

The scientists started by analysing 150,000 Facebook photos, each of which had been shared at least five times. In order to protect privacy, the data was stripped of names and identifiers.

According to a preliminary analysis of the photos, at any given point in a cascade, there was a 50-50 chance that the number of shares would double.

The researchers then looked for variables that might help more accurate prediction of doubling events than a coin toss, including the rate and speed at which photos were shared, and the structure of sharing (photos reposted across multiple networks proved to create stronger cascades).

After taking several factors into their analysis, the computer scientists were able to accurately predict doubling events almost 80 per cent of the time.

The accuracy of their algorithm increased the more number of times a photo was shared and for photos shared the accuracy rate approached 88 per cent.

The speed of sharing was the best predictor of cascade growth and simple analysis of cascade unfolding predicted doublings 78 per cent of the time.

According to Leskovec, slow, persistent cascades did not really double in size.