My first encounter with metadata was 2021, during the pandemic and the rise of CNN in SoC that performed much more efficient than machine learning. In short, the metadata in video surveillance focused on people and vehicle (often called ‘objects’): detecting and not detecting, the path, the appearance color, facial attribute, etc.
All in all, trying to provide better answer to:
- More accurate alarming
- Faster video search speed
- Better clarity of people and vehicle
In SharePoint CMS project, I had the chance to look at metadata from a different angle: documents.
I think at the core, it’s locating the file faster through data and visual presentation.
- Layer 1: user defined metadata such as person in charge, product release date, budget amount, etc.
- Layer 2: assign metadata to the documents. Ie, file, folder, or document set.
- Layer 3: apply in visual presentation such as columns, filter in web parts, creating views, etc.