Predictive Analytics: Using Twitter Data (Part 2)

Extracted 15NOV2011 from http://online.wsj.com/article/SB10001424052970204138204576598942105167646.html#printMode

In July, the Pentagon's Defense Advanced Research Projects Agency began exploring the possibility of a $42 million effort to detect online "persuasion campaigns" and "influence operations" aimed at spreading ideas through Twitter and other social media...

The new messaging medium has also spawned a new form of political deception, in which campaign operatives marshal an array of dummy Twitter accounts to spread rumors or misinformation. Like form letters, robo-calls and push polls, these Twitter tactics are inexpensive, since user accounts are free, and can potentially reach many more people than traditional campaign attack ads...

...researchers at Indiana University have been working on ways to detect and defuse Twitter misinformation campaigns automatically. But the technology of Twitter is moving so quickly that detection efforts can barely keep pace. "People can game these systems and, in gaming them, they help bias the results of any data company," said social media analyst Danah Boyd at Microsoft Research. "It's a real challenge."

Pitting machine intelligence against human gullibility, researchers at the Web Ecology Project in San Francisco are using Twitter as a proving ground for advanced pre-programmed personalities called "socialbots" that can engage in extended conversations via Twitter by imitating the behavior of real people sending and receiving messages...

Network sociologists are worried that these newest contrivances may offer others a powerful way to manipulate people through Twitter on an even larger scale. "Doing this on Twitter with a thousand accounts or a million accounts is the next step," said Indiana University computer scientist Jacob Ratkiewicz.