If you’ve ever searched for something and seen a panel to the side of the main interface that displays some facts related to your query, then you’ve seen the Knowledge Graph in action. The company first introduced the database back in 2012. Since then, it has expanded to include approximately 500 billion facts related to 5 billion entries.
As you might have guessed, ensuring Search connects an image with the correct facts can be tricky. Google says it’s using a deep learning algorithm to parse and evaluate an image’s visual and text signals. It then combines that with its understanding of the text on a website. The company says that information helps it determine the most likely people, places or things related to a specific picture.
Currently, this latest extension of the Knowledge Graph is only available on mobile devices in the US. Moreover, not every image you tap on will include related information. That said, Google plans to expand the feature to make it compatible with more languages and pictures over time.