The Library will retire the Newspaper Navigator application on April 21st, 2025. Created by Benjamin Charles Germain Lee while he was in service as a Library of Congress Innovator in Residence, the application has received over 174,000 visitors representing tens of thousands of research experiences.
Ben’s project was the first in-house machine learning application developed at the Library. It used machine learning techniques to extract visual content from over 16 million pages of digitized American newspapers in the Chronicling America collection. Researchers could browse and discover images from the collection by search term and further refine their searches by visual similarity.
While the time has come to bid this experimental application goodbye, the Library continues to learn from and build on this work. We know how much Newspaper Navigator has meant to us, and we would love to hear from past users about their experiences with Newspaper Navigator. Please share your story with us in the comment section below.
As part of our goodbye celebrations, Ben Lee had some parting thoughts around the application, its usage and the range of projects and use cases it has supported over the years.
Note: Privacy and publicity rights for individuals depicted may apply.
“When I began Newspaper Navigator over five years ago, my goal was to reimagine search and discovery for the incredibly varied and rich visual culture found within the millions of newspaper pages within Chronicling America. Since then, the resulting Newspaper Navigator dataset has supported millions of computational queries, and the search application has supported tens of thousands of individual user experiences.”
He added, “I have been amazed by the inventive ways that the public has used Newspaper Navigator. Teachers have used the project with students in the classroom to teach subjects as far-ranging as social studies and computer science; genealogists have utilized the project to find photos of family members; scholars have used the dataset to study photographs and editorial cartoons, as well as advance research within AI. It has been a highlight of my career to work so closely with LC Labs, Serials, and IT Design & Development at the Library of Congress. It was only with these collaborators that Newspaper Navigator was possible, and I am forever grateful that we were able to reach so many members of the American public.”
If you’re curious to learn about the impacts of the application and the relationship between the Chronicling America collection and the Newspaper Navigator data, the following white papers offer a deep dive: Compounded Meditation: A Data Archaeology of the Newspaper Navigator Dataset and The Newspaper Navigator Dataset: Extracting and Analyzing Visual Content from 16 Million Historic Newspaper Pages in Chronicling America. While the application itself will be retiring, we aim to make the Newspaper Navigator datasets publicly available and free to use for future projects, experiments and ideas.
For more information about the application, please visit the Newspaper Navigator page on the LC Labs website and continue to stay tuned to the Signal for more updates.
Comments (9)
Newspaper Navigator is a joy. It is like finding an unexpected door to a long corridor of surprises, every new door a revelation.
McLuhan put it best: The medium is the message. Especially true of newspapers!
Newspaper Navigator revolutionized the way that we think about searching newspapers. Ben Lee’s innovative use of existing data to create a whole new tool speaks to what is possible when researcher engages not only with a library collection, but with the librarians who created that collection. I hope that the foundation Ben Lee has laid here will lead to further exciting ways to navigate the ever expanding and always relevant collection that is Chronicling America.
This took gave me a new way to search for content, and it has provided matches that text search did not. It is a loss for researchers that this tool will be discontinued. Please consider building an updated image-search tool soon.
This tool gave me a new way to search for content, and it has provided matches that text search did not. It is a loss for researchers that this tool will be discontinued. Please consider building an updated image-search tool soon. I will look forward to it!
5 years, 174,000 users – that’s almost 100 per day. And while the image search database will be preserved, I hope that the user functionality of performing an image search through the Chronicling America collection will also be preserved.
Excuse my ignorance, but what is NN’s replacement? Ai?
Thanks for the reply.
Thank you for your comment, Kitt. While Newspaper Navigator was always meant to be a temporary research experiment, Chronicling America is the source of the data behind NN and it’s available better than ever and with a commitment to continuing. If you’d like to continue exploring this data set, another cool tool is Exploring America’s Newspapers, an application that maps all the newspapers in the Chronicling America collection.
Frankly I do not understand why you are retiring a fantastic resource, it looks as the usual tech “update” operation whereby, instead of improving things which are already quite good or even excellent, they are idiotically degraded.
Thank you for your comment, Daniel. We also love Newspaper Navigator, and it was developed as a temporary research experiment to explore what’s possible. The data behind Newspaper Navigator has grown well beyond the original size of the date set (Chronicling America has expanded from 16 million to 22 million pages), so we need to look for new ways to actively apply the machine learning pipeline developed by Ben Lee. In the meantime, another interesting way to explore this data set is through Exploring America’s Newspapers, an application that maps all the newspapers in the Chronicling America collection.