Researchers Alok Choudhary, PhD, and Ramanathan Narayanan, PhD candidate, created pulseofthetweeters.com, to measure Twitter users’ influence.
Having thousands of followers on social networks doesn’t guarantee great influence. While a celebrity may have thousands of Twitter followers, it doesn’t mean people are taking his tweets seriously, says a study from Northwestern University examining social media and influence.
The purpose of the new site, which uses a specialized algorithm to determine who the most influential tweeters per topic are, is to narrow down and specify information. Combining dynamic data mining, sentiment analysis and network analysis in real time, the algorithm identifies the most influential tweeters, and classifies their tweets as positive, negative or neutral. It also offers related topics for visitors to read about.
"There are about 50 million tweets produced every day, but most of us only read 10 or 20 tweets in one sitting," Narayanan said in a public statement. "So, which tweets should you read? Which tweets are being read by media experts on any given subject, such as politics, law, fashion, food? We provide that information for users."
Choudhary, who is the chair and John G. Searle Professor of Electrical Engineering and Computer Science at Northwestern, has been doing data mining for a decade. He tellsdemodirt.com that today people are bombarded by so much information that it has become a challenge to navigate through it all.
“In any domain, when you have a large number of people, there are advertisements, blogs, television, newspapers—whatever it is—there is a huge amount of data, and it is not possible manually to figure out what is going on,” he explains.
In the millions of conversations around the world, Choudhary notes, there are “microclusters” in which people discuss numerous topics. “In these microclusters, how do we know who is influencing who?” he says.
Moreover, does having thousands of Facebook fans or Twitter followers mean that your opinion has any influence over the masses?
No, Choudhary says, adding that it depends other factors.
“There are many different arenas [of information],” he maintains. “An individual user may follow different people—some [experts] from sports, some from politics, [and so forth].
People do not follow one person as an expert on all topics, Choudhary adds. For example, a talented basketball star’s opinion of a Supreme Court justice is less likely to have any influence over his followers, as his comments about a sports-related issue would, he explains.
The main advantage of this new development in data mining, he contends, is the ability to discern between influencers according to topic.
"Discovering patterns, opinions and sentiments from a massive number of tweets is challenging in itself, but discovering influencers and leaders for specific topics is a major technological advance in data mining," Choudhary noted in a public statement.
With dynamic, drastic changes in how we communicate, online social media must constantly advance as well.
“In this new social world the way people make decisions and get influenced and take action is not like what used to be in broadcast type of media or in a static, universally-applied way,” Choudhary explains.
“Thirty, forty or fifty years ago, Walter Cronkite would be influential, but today people have access to information and are part of their communities, of various communities,” he adds. “There are more dynamic measures that one would take into account. There are lots of different virtual communities, with similar passions, and by taking static measures it is impossible to capture them.”
Read this article at demodirt.com
Written by Galia Myron Thursday, 30 September 2010 12:45
| < Prev | Next > |
|---|
