If you’re not familiar, Pandora is a Web site that takes a visitor’s preference for an individual musician or song and creates a personalized radio station to fit his or her taste. If you like the Beatles’ “Paperback Writer,” you may also like “Ruby Tuesday” by The Rolling Stones, for instance, or “I Can’t Explain” by The Who. With Art.sy, a visitor can enter an artist, artwork, artistic movement or medium into a search bar and the site will generate a list of artists and works that have been deemed related in some way. “There are a lot of people who may know who Warhol is, but they have no idea who Ray Johnson is. The ability to make those connections is what this is about,” said Cwilich, Art.sy’s Chief Operating Officer, on a recent segment of The Takeaway with John Hockenberry.
The endeavor is a true collaboration between computer scientists and art historians. (This is even evident in Art.sy’s leadership. Cleveland, Art.sy’s 25-year-old chief executive officer, is a computer science engineer, and Cwilich is a former executive from Christie’s Auction House.) To create a Web site that could generate fine-art recommendations, the Art.sy team had to first tackle the Art Genome Project. Essentially, a number of art historians have identified 800-and-counting “genes,” or characteristics, that apply to different pieces of art. These genes are words that describe the medium being used, the artistic style or movement, a concept (i.e., war), content, techniques and geographic regions, among other things. All the images that are tagged with a specific gene—say, “American Realism” or “Isolation/Alienation”—are then linked within the search technology.