When it comes to the growth of data, there isn’t much that can compare to the explosion of online video.
YouTube, for example, says its users upload more video in a month than NBC, CBS and ABC created since they first started broadcasting in the mid 1940s. Reporters and audiences alike can now capture and share newsworthy events with near ubiquitous technology.
But aside from text like titles, keywords and descriptions, all this data is hopelessly opaque to analysis: unsearchable, unindexed and unstructured.
That’s a big problem for any journalist trying to make sense of it all. And it’s already led to a few novel solutions.
The Knight Foundation and Mozilla, for example, announced they’re pumping $1 million into a start-up called Amara, which provides a platform for users to caption and translate online video. Here at the lab, we’re working on our own annotation tool for video and audio, called Video Notebook, which allows users to mash up social media with video and add their own notes. To add a clickable index of speakers, bills and other information to Virginia General Assembly video on Richmond Sunlight, Knight News Challenge Fellow Waldo Jaquith hacked together a system for processing the chyron text that pops up on the screen.
Adding structure in that last example – government video or audio created as a matter of public record — is of particular interest to reporters like Liam Dillon, who covers city government for the Voice of San Diego.
He typically attends local meetings two to three times a month, capturing everything on an audio recorder. And that can mean hours of playback time later.
But when he’s looking for something specific from a past meeting, Dillon occasionally employs another tool designed to help him and other San Diego citizens sift through and locate points of interest in hundreds of hours of government video.
Granicus, a vendor that agencies like the city of San Diego pay to help manage their recorded public meetings, has built tools to help recordkeepers tag, index and caption video on the fly. And with expanded tools on the horizon, the company’s technology could become even more valuable reporters and developers looking to keep tabs on elected leaders.
A natural fit
Granicus began as a general video platform, but Chief Technology Officer Javier Muniz said he and his fellow co-founders soon recognized the need for long-form video suitable for lengthy government meetings. And their platform was already designed to handle 8 to 10 hours of content.
Tools for reporters
Global search function
While each government agency’s Granicus portal allows users to search within its own archived content, the team also developed a global search feature.
Some videos in the Granicus archives allow users to select and embed video clips of government meetings and legislative sessions using code similar to YouTube.
Granicus’ open API allows developers to tap into part of the company’s archive of government data. Registration is required.
“We kind of saw the writing on the wall that government was going to have to transition to streaming video on the Web,” Muniz said.
But the team also wanted to do more with the video content their clients were creating — namely to match it up with the detailed agendas, voting records and transcripts clerks were already logging, often while meetings and sessions are in progress.
“They’re already there recording data in a really organized fashion that can be repeated in software,” CEO and co-founder Tom Spengler said. “The whole time, they’re indexing.”
By giving government employees the option to integrate this content into video as it’s happening, they don’t have to do it after the fact.
“No one wants to watch a city council meeting twice, especially when you were in it,” Spengler said.
The result is a collection of videos posted within hours that are already searchable by agenda items — and depending on the software purchased by the agency, closed captioning and voting records as well. A pulldown menu also allows users to jump to the specific agenda items within the video itself.
“That’s particularly helpful with a discussion topic that takes a few hours,” Dillon said in a March phone interview.
The Granicus team is already working on additional ways users can work with the data tied into its video archives.
A search feature, more or less in beta at the moment, can probe keywords in the agenda items stored across every agency (voting records, captions and the like aren’t searchable yet). Users can also search specific agencies to find their most recent videos. Because they’re still testing it, Muniz said its results are usually about a day behind.
They also have an application programming interface for developers looking to play with the data itself. Sunlight Labs even put it to use during its 2011 Labs Olympics, which resulted in a Granicus-powered project called Talk of the Town.
The API and search functions are still in development, but Muniz said they may divert more resources to the features if they become more popular.
Muniz said his team’s next step is to look into adding functionality like real-time topic monitoring and alerts for those who want up-to-the-minute notifications for specific government activity. There’s no timeline yet for when that will be released.
One noticeable drawback: Granicus is a bit slow when compared to other common video hosting solutions. Jumping to different agenda items, for example, means extra buffering time (Muniz says Granicus is moving away from Microsoft’s Silverlight video player, which is causing the slowdown). Users like Dillon say that it’s not a huge problem though.
“This is the sort of minor stuff that could be improved if it was a more journalist-friendly tool,” he said.