Yesterday SMART@znmeb (SMART stands for “social media analytics research toolkit”), a SUSE Linux appliance created by Ed Borasky, added sentiment analysis to its set of features. The toolkit now includes texttir, a sentiment analysis package created in the statistical programming language R. SMART@znmeb includes other open source tools that include data mining, dashboarding and data visualization.
Borasky says textir is the first open source sentiment analysis library he’s found that he thinks may actually work. “Most of the vendors sell a sentiment analysis tool of some kind or another, and the customers that have tested multiple tools spend a lot of time trying to figure out why they give different answers,” he says. He also cautions that sentiment analysis is vulnerable to spam and other gaming tactics and requires a large investment in hardware.
Borasky also added textir to Project Kipling, another SUSE appliance he designed for data journalism.
Borasky created SMART@znmeb at Open Source Bridge in Portland in June 2009. “It was a spur-of-the-moment thing,” he says. “I was initially looking for a way to combine my two main interests at the time, Linux and Twitter.” @zmeb is Borasky’s Twitter Handle. “A few weeks later, the SUSE Studio project opened up publicly and that was the final push – I could package Linux software into appliances on line, for free,” he says.
SMART@znmeb is a great case study in how off the shelf free open source software can be applied towards business problems. It can be used as a replacement for or in conjunction with many business analytics and social CRM tools. “I’ve done a few tests of that nature, interfacing it with the Viralheat and Clicky APIs and with Twitalyzer CSV reports, but I built those mostly for myself to help manage my blog,” Borasky says.
SMART@znmeb users do need to have some background with Perl, R and PostgreSQL to use it.
At the moment it’s a non-commercial project. It doesn’t cost Borasky much to maintain since all the software is free and SUSE Studio lets him distribute the package for free. But he says he’d sell applications built from the tools or consult on applying the technologies.
Photo by kirinqueen