������̳

Converge Media is an Emmy-winning video producer based in Seattle. For 10 years, it has been telling and uplifting the stories of the Black community in the Northwest. 

In that time, it has collected a massive archive of videos and photos — 160 terabytes. But when Converge was first beginning its work, it didn’t have a great system for labeling its content. It has a wealth of content featuring members of the Black community. What it doesn’t have is a great way to find that content. 

A group of University of Washington Information School students are working to help solve that problem for their Capstone, a culminating project that gives students a chance to tackle a real-world challenge. 

The students, Zach Greenman, Skye Shen and Kexuan Feng (pictured at top, from left), along with Shivang Agrawal (not pictured) are all earning their Master of Science in Information Management degrees. The system they created uses artificial intelligence to help identify people featured in Converge’s archive. Their work will make it dramatically faster and easier for Converge to find old photos or videos featuring specific people. 

“Converge Media is local and they really care about Black people’s lives,” Shen said. “The first time we met, they told us that our tool is going to be an actual tool that they will use in the world. It was going to have a real-world impact, and that was very fulfilling. That’s a thing we wanted to contribute to.” 

The system the students created will eventually be able to process all of Converge’s archived media and tag the people who are featured. The process will take time because there’s so much archived content. Once it’s done, Converge will have an index that will make it much easier to find old information. And, as the company adds new content, the system can tag it and update the index. 

When the students began working on the project, they talked to Converge about the company’s needs. They quickly realized that, as a small company, Converge needed a simple system that didn’t require expensive or complicated technology. The team ultimately decided that the best choice was to keep all the workspace within Google Drive, which Converge staff already use. The system maintains the staff’s existing folder structure and working style. 

“We had considered a database,” Greenman said, “but folks already know how to use a spreadsheet. And it works for the case that we have. We don’t need a more technical solution. We tried to make it grounded. It’s not fancy, but it’s very useful.” 

Another important consideration for the team was how facial recognition functions, especially given concerns that it can perform differently on Black faces than white faces. The team reviewed literature to give them a deeper understanding of the issue. To help Converge identify faces, the tool goes through multiple steps. First, faces are located, but not identified. Then, faces that are likely the same person are grouped together. The actual process of connecting a face to a name is done by Converge staff. That means the archive becomes searchable by name in Google Drive, while keeping a human reviewer in control of identity decisions.

“We did not want the system making confident identity claims on its own, especially for an archive centered on Black community stories,” Greenman said. “The goal was to make Converge’s review process faster and more systematic, not to replace their judgment or community knowledge.”

The team members said their education at the Information School gave them the tools they needed for this project. And, when they came across something they didn’t know how to do, they knew how to find a solution. 

“The ������̳ program makes you versatile,” Agrawal said. “You are not framed into being just a business scientist or a data analyst. That makes it interesting for me. It allowed me to do things I never would have tried.” 

In particular, Agrawal enjoyed working on user experience. The team consulted with Converge staff about their needs throughout the project and devised a technological solution, but one that could be used by staff without a lot of technological knowledge. 

Feng also appreciated the chance to work closely with the Converge team and do a lot of user testing — and to create a product that will immediately be helpful. 

“We are able to really help make the archive easier to search and use,” she said. “They have years of local Black stories and community moments. And this will help them reconnect to stories and memories.”

Omari Salisbury, the founder of Converge, said the new tool will help to make the most of the power and potential of the company’s archives.

“It will unlock so much for the future of our storytelling,” he said. “It can help us give readers and viewers context today for what was done or said in the past.” 

The timing is good, too, because the students’ work will help support storytelling for Converge’s 10^th anniversary. Without this tool, it would have been too time consuming to look through the files for the content they needed. 

“I think the biggest strength that this group has is imagination and creativity,” Salisbury said. “And they have a much better picture of where AI is than I probably ever could. They were able to meet us where we’re at. We don’t know all the funny AI terms. We just know what we envision or what we imagine. So, for them to able to work with us and across disciplines, it’s really dope.”