Stephen E. Arnold: Mining [Big] Visual Content [Data]

Advanced Cyber/IO
Stephen E. Arnold
Stephen E. Arnold

Mining Visual Content

Data-mining of social media is moving beyond natural-language processing of Tweets and Posts and into a whole new realm of visual content interfacing to analyze the data available on the heavily-visual social network Tumblr.

Cierra Buck posted “Introducing Tumblr Firehose data on the Pulsar Platform and a whole new interface for mining visual content” on the Pulsar Platform Blog. In the post, Buck notes that Pulsar integrated Tumblr not just as a blogging site, but as a social networking site. But because over 80% of Tumblr content is visual, integration required an entirely new way of presenting data to Pulsar platform users:

“This allows Pulsar to display the actual image and video content rather than a preview end enabling endless scrolling rather than organising the content in pages. This allows for easy browsing of rich media social content which, coupled with advanced filtering using all the metadata we generate, is going to give you a powerful mining tool to uncover visual patterns and trends in your dataset.”

Buck’s post gives a good visual tutorial of how the interface will work and the kinds of data it will analyze from Tumblr. Do the kids on Tumblr realize that everything they post is being mined?

Laura Abrahamsen, February 23, 2014