Twitter just bought a machine learning company that uses software to extract data from images, which means it knows what Facebook, Tumblr and Pinterest know: Pictures are important.
With Madbits technology, Twitter will be able to glean information like metadata and image content from pictures shared on the social network, and in turn figure out what’s visually popular and why.
We’re already sharing 140-character blasts of information on Twitter, and it’s fairly easy to figure out what people are talking about by analyzing text. For instance Twitter’s “firehose” of data provides a handful of companies access to all the tweets and activity streams going back to 2006. With his information, people can discern what was popular at any given time. It’s much harder, if not impossible at this point, with images.
“We developed our technology based on deep learning, an approach to statistical machine learning that involves stacking simple projections to form powerful hierarchical models of a signal,” the company said in a statement posted on its website.
Twitter has made significant strides to improve visual aspects of the social network. Thanks to explosive popularity of image-heavy sites like Pinterest and Tumblr, social networks that rely on text have had to play catch up on the Visual Web. Twitter has been relatively slow to adapt, but implementing features like in-line photos and, more recently gifs, show it’s focused on being a visual network, and giving photos as much importance as words.
And, as a result, figure out what humans want to look at.
Lead image courtesy of Camera Eye Photography on Flickr