Disillusioned by AI indexing? Newsbridge is taking it beyond a bundle of tags with Multimodal Rules
My brother Philippe and I founded cloud media company Newsbridge in 2016, driven by a desire to help make audiovisual content live forever. It’s a mission that we share with FIAT/IFTA, and is the reason why Newsbridge is a Gold Sponsor of the upcoming World Conference in Cape Town.
Newsbridge wants to help organizations immediately access all the content they’ve ever produced. We promise that our users will be able to find the photo or video clip they need in less than 2 seconds, whatever the size of their dataset. We could be talking about half a century of news archives, 200,000 hours of footage or 40 live, simultaneous video streams.
Central to Newsbridge’s media hub solution is our AI Indexing technology. Before you make that big sigh, rest assured, we’re right there with you. We understand the deep disillusionment that many archivists feel about artificial intelligence. Every two years, the volume of content produced worldwide doubles. What was touted as a revolutionary solution for broadcasters seeking to keep up and archive photo, video and audio content at scale has, for many, resulted in a big data indexing problem. Documentalists find themselves with an overwhelming amount of data to handle. A huge number of tags have been generated, but all that tagging proves impossible to decipher when the AI is not speaking your organization’s language.
Making real sense of your metadata
What sets Newsbridge’s AI apart are our Multimodal Rules. Multimodal AI is based on data fusion. It means we are merging detection results from face, text, objects, pattern, context and transcription. All of this contextual information produces incredible metadata. Combined with our thesaurus and semantic search engine, Newsbridge is able to detect advanced editorial concepts and increase search accuracy at scale.
Mono or unimodal AI may index, for example, interview footage by detecting a table, perhaps the microphone and water glass placed upon it. It will identify the people or person talking, and the fact that they may be wearing glasses. But 99% of AI indexing tools on the market today won’t produce the tag “press conference” which is what users will want to type into a search bar when looking for the footage.
Multimodal AI powering multilingual news
One of the sessions I’ll be co-presenting in Cape Town is with Newsbridge customer Asharq News, a 24/7 multi-platform news service. Asharq News is taking full advantage of the customizable indexing models offered by Multimodal Rules, and is training the AI in its organization’s language. Literally, this is Arabic and English. A highlight of its indexing model is the ability for users to search through metadata in one language to find content tagged in another. Newsbridge researchers are making new discoveries about our Multimodal AI’s capabilities all the time. If you want to know how early fusion is being used to automate the detection of editorial highlights, and the next big thing in video indexing, you’ll want to make sure you catch my conference presentation with Asharq.
Newsbridge is passionate about accelerating media production workflows and providing easy access to content. It’s why we created our Smart Collections feature. Once tags are set, these dynamic folders will automatically update with any future clips ingested to the Newsbridge Media Hub that match the filters – no manual sorting required. Individual collections can be securely and easily shared with partners and external buyers. The time saved gives teams the bandwidth to focus on higher-value tasks.
Leading the archive evolution
I’m often asked what AI means for the role of archivists in the future. As well as continuing to be content guardians, I believe they’ll play an important role in harnessing the power of AI and molding it to meet specific content indexing and archiving needs. Deeper learning leads to more accurate results, and ongoing training is needed for AI to precisely identify the key people, objects, locations, context and actions that matter to your ever-evolving organization.