Who gets to determine the big topics of conversation on social media? And how do they do it? Looking to find out, HP researchers recently examined how popular subjects get to be listed among the top ‘trending’ topics on Twitter.
“You might expect the most prolific tweeters or those with most followers would be most responsible for creating such trends,” says Bernardo Huberman, HP Senior Fellow and director of HP Labs’ Social Computing Research Group. But that turns out not to be the case.
In a new paper, Huberman and three fellow researchers demonstrate that “user activity and number of followers do not contribute strongly to trend creation and its propagation.”
Instead, says Huberman, “we found that mainstream media play a role in most trending topics and actually act as feeders of these trends. Twitter users then seem to be acting more as filter and amplifier of traditional media in most cases.”
How trends start
The main determinant of whether an item trends – much more than who tweets about it or how often – was the specific subject of the tweet, the team found. That can be seen in the degree to which trending topics are the result of retweets – items passed from one individual’s network of followers to another’s. 31% of tweets of trending topics are retweets, their analysis showed.
The role of “traditional” media sources
The HP team collected data from Twitter’s own search API over a period of 40 days in the fall of 2010. From the resulting sample of 16.32 million tweets, they identified 22 users who were the source of the most retweets when a topic was “trending.” Of those 22, 72% were Twitter streams run by mainstream media outfits such as CNN, the New York Times, El Pais and the BBC.
Similarly, the research showed that just having an active Twitter account was not a factor in creating a trend.
Predicting the ebb and flow of trends
Huberman and his colleagues also wanted to know why some trends last longer than others. Very few topics stay much more that 40 minutes at the top, they discovered, but those that do persist engage an especially diverse audience. Indeed, says Huberman, “we showed that the distribution of long-time trends is predictable, as is as the total number of tweets and their growth over time.”
Crowdsourcing the public agenda
Even though mainstream media sources tend to be extremely well-represented within social media discussions, a significant percentage of trending topics do stem from non-mainstream sources, Huberman notes. And that’s a change from the pre-internet era when social media consisted of little more than conversations held in person and over the phone.
However they originate, though, the fleeting nature of trending social media topics means that they’ve yet to show they can really influence the public agenda. “In traditional media, being on the news for several days is enough to start a conversation about that topic,” explains Huberman. “In social media, a few hours might do as well, but we have no evidence yet that it is the case.”
Download the paper at Scribd.com here
Why HP researches social computing
Understanding how social media might be altering the nature of information transmission and social influence is a major goal of HP’s Social Computing research. To that end, its Social Computing Research Group is creating novel tools to analyze the flow of “unstructured data” (such as social information) and developing new services that dynamically respond to the ways in which information is now being created and consumed online.
Other areas of focus for HP’s social computing research include harvesting the collective intelligence (i.e. the “wisdom of the crowd”) and designing incentives for accessing and allocating resources in the most efficient ways possible.
More HP social computing research
Image credit: pdinnen