The Evolution of Active Archive with Machine Learning

In recent years the term “active archive” has been a frequently referenced use case and its granular power and value has, in my opinion, been somewhat overlooked. 

For me, a cloud-based archive is a service where client data sets are stored in a vendor’s infrastructure such as Wasabi using web-based tools, and access to data gets moved from on-prem, in-house systems to a cloud destination in the chosen vendor’s network.  

The “active” element refers to data stored in a sufficiently performant service that allows curated data to be searched by metadata, located, chosen, and retrieved from the cloud service to local systems for manipulation or re-use.  

When a user has hundreds of thousands or even millions of digital assets, interrogating the datasets that contain them to access and unlock its value is potentially hugely challenging. It’s data on a scale no human—or team of humans—could ever address. 

But there are many powerful, granular cataloguing and indexing tools available that take much of the grunt work out of managing such large swaths of data.  

Machine learning tools have become extremely adept at identifying visual and auditory elements in video. They can be applied to convert speech detected in audio clips into searchable text, or tag videos with appropriate metadata for more efficient indexing. The smarter solutions can be trained to apply facial recognition to enable historic or esoteric content to be searched by the image of an individual.  

Baseball is a pretty big deal to Wasabians (especially in the U.S.) and the ability to detect teams and brands by their logos is now easily achievable with the aid of ML tools, bringing with it many associated value-streams. For example, highlight packages can be assembled in a breeze with the proper keyword: a player, a stadium, a date.  

Artificial intelligence and machine learning technologies are redefining what we can do with our active archives. By quickly indexing an entire media library, organizations can extract new value from their existing content. 

Please stop by the Wasabi booth at NAB in April to see a brilliant example of this technology in action from the fantastic GrayMeta whose brilliant Curio product is the inspiration for this short piece.