Teenage social steganography and sentiment analysis

Social Cockpit Dashboard Measuring Social Media

Interesting thoughts from Danah Boyd (@zephoria) about how teenagers are using social media. It makes we wonder about the accuracy of “sentiment analysis” software that’s used by large organisations to derive meaning from social networks. Danah:

“Pew’s report shows an increase in teens’ willingness to share all sorts of demographic, contact, and location data. This is precisely the data that makes privacy advocates anxious. At the same time, their data show that teens are well-aware of privacy settings and have changed the defaults even if they don’t choose to manage the accessibility of each content piece they share. … This is the practice that I’ve seen significantly rise since I first started doing work on teens’ engagement with social media. It’s the source of what Alice Marwick and I describe as “social steganography” in our paper on teen privacy practices.

“Over the last few years, I’ve watched as teens have given up on controlling access to content. It’s too hard, too frustrating, and technology simply can’t fix the power issues. Instead, what they’ve been doing is focusing on controlling access to meaning. A comment might look like it means one thing, when in fact it means something quite different. By cloaking their accessible content, teens reclaim power over those who they know who are surveilling them.”

via danah boyd | apophenia » thoughts on Pew’s latest report: notable findings on race and privacy.

How accurate are social media sentiment analysis tools?

UK police using are Radian 6 as a listening tool and for trend analysis, and RepKnight to identify geolocation hotspots…

“Umut Ertogral, who runs the Opensource Intelligence Unit for London’s Metropolitan Police Service, today told the AusCERT information security conference a team of 17 staff were working seven days a week to track social media feedback and monitor community tension.

“There’s a lot of work we’re doing to analyse the language and how people are talking on Twitter,” Ertogral said.

“Companies will tell your that sentiment analysis from a piece of software is about 56 percent accurate … we would say it’s lower, because it doesn’t pick up humour or slang,” Ertogral said.

/via http://mobile.itnews.com.au/

Photo via Intel Free Press