Library of Congress Looks to AI to Help Users Sift Through Its Collection

The library has chosen three computer science and digital humanities experts to investigate how branches of artificial intelligence such as neural networks and computer vision can help humans sift through some of its millions of digitized items. The Morning Download delivers daily insights and news on business technology from the CIO Journal team. "We are all inspired by the technology we see in our everyday lives," such as photo apps that use facial recognition to group together pictures of the same friend or family member, said Kate Zwaard, director of digital strategy at the Library of Congress in Washington, D.C. "People are really interested in applying that technology to the [library's] collections." The library's current search technology relies on the indexing of text and metadata records that allows items to be recognized by a computer. But there are millions of items that cannot be found through search, in part because the task of categorizing each item is impossible for humans to do alone, Ms. Zwaard said.