Analyzing Social Media: Graffiti and a Tweet Heard Round the World

6/25/12Follow @bvbigelow

(Page 2 of 3)

controlled faucet. Now it’s more like a neural system that reacts to different kinds of stimulation.

Giblin says this line of thinking spurred the development of Graffiti—a software-as-a-service system that the news media could use to sharpen the focus of its content—and attract bigger audiences—by making better use of comments scattered across a variety of social media platforms. “We can take the comments on every article, image, video, and music file, read it, and apply artificial intelligence, machine learning, root word analysis, latent semantic analysis,” Giblin says. “We crunch it down to a CliffsNotes version for anyone who wants it.”

Giblin says the distilled product of Graffiti’s analysis consists of the most relevant keywords and terms from social media arising from specific events or content—and the system can then update the content by inserting these keywords as the metadata tags used by Internet search engines. In essence, the process serves as a kind of feedback loop for search engine optimization (SEO).

“The Miami Herald has no idea about what people are talking about on Facebook, but we can tell them, and they can use that information to focus on the core issues that people are most interested in,” Giblin explains.

For example, a Graffiti analysis of social media commentary about the fatal shooting of Treyvon Martin might show that people rarely mention the victim’s name, but frequently use other terms—like “hoodie”—in their online discussions. People don’t talk about something with proper names and titles, Giblin says. They say “that hoodie killing in Florida” or “that plane that went down on the Hudson.” “Hoodie” would therefore be a more important keyword for … Next Page »

Bruce V. Bigelow is the editor of Xconomy San Diego. You can e-mail him at bbigelow@xconomy.com or call (619) 669-8788 Follow @bvbigelow

Single Page Currently on Page: 1 2 3 previous page

By posting a comment, you agree to our terms and conditions.

  • http://twitter.com/tom_semantic Tom Marsh

    Great explanation and article on 451′s value prop vs other sentiment analysis applications.  We’ve known Patrick and his team for awhile and they are going to carve out a big part of this market once they get fully commercial with the platform.  Good to see his story picked up.