We are on the cusp of a revolution in content discovery, one that will result in content catalogues that are easier to navigate, content descriptions that are more accurate and valuable, search that is far more granular and recommendations that are made more useful by taking account of factors like plot and mood. The overall impact will be to make TV more personal than ever without sacrificing the serendipity that entertainment lovers value.
This generational upgrade in content discovery is underpinned by a second revolution – this time relating to metadata technology and processes. Having always been one of the less fashionable topics in television, metadata could be about to take centre stage, raised to C-level discussions because of its new business implications.
Perhaps the most exciting development is ‘augmented metadata’, which relies on the ability to generate new metadata by analysing the video scenes using image recognition and the closed-captions within content. Image analysis, which is possible on a frame-by-frame basis if necessary, includes facial recognition and setting analysis. This is made possible thanks to machine learning. For every scene, you can determine who is in the scene and where they are.
Scene analysis is the basis for deep-search capabilities. With augmented metadata, viewers can search for an episode of ‘Modern Family’ where the character Jay was on a golf course. This is only possible by using a combination of facial and scene recognition. The use of subtitle analysis makes even more complex search requests possible, such as: “I want to find the scene in the ‘Big Bang Theory’ where Sheldon and Leonard are in a lab talking about the Space Time Continuum”.
A White Paper published this week by Piksel (which designs, builds and manages online video solutions and has now announced its Fuse Metadata solution) demonstrates how content themes are identified as a way to improve user discovery. Themes become one of the building blocks for what the company calls ‘video DNA’, which refers to the underlying character of the content plus what happens within a show or movie.
Following contextual analysis, the metadata file will be able to list the main themes within the content, the dominant feelings and even the personality traits of different characters (based on language analysis). Plot-DNA analysis can outline what characters do and what they feel in any video frame. As an important ancillary benefit, a knowledge of what is happening at any given moment within content creates opportunities for more contextual advertising.
Metadata augmentation leads to the creation of micro-genres, so that instead of presenting a movie as a ‘thriller’, it can become a ‘black-and-white thriller from the 1940s’. Instead of presenting ‘horror’ as a search option, users can look for ominous, exciting or suspenseful horror.
Augmented metadata can be created on-the-fly to extract information about who is in a scene and what they are talking about, among other things. Linear television including news, financial news and sport can be chaptered, creating on-demand files that reference subjects that may be of interest to consumers, with start and finish time-stamps.
The White Paper outlines the problems with metadata management today and how this can be changed in order to improve the business prospects for service providers, broadcasters and other content owners.
A metadata consolidation process analyses each metadata file that arrives alongside video content and seeks to match it against the existing metadata catalogue. If the content exists already and there is metadata for it, duplication is eliminated. Parts of the new metadata can be merged into the existing metadata or the new file can replace the old (or the new metadata is ignored altogether).
A single master file is maintained with different language options and the cover-art and graphics that are suitable for different markets. This master file also serves every platform and end-device. It can contain different versions of the cover-art and synopses to suit the device on which the content will be viewed. Thus a laptop will access a longer movie synopsis than a set-top box, and an Apple TV will fetch higher resolution artwork than a smartphone.
According to the White Paper, there is no more title duplication in catalogues when using next-generation metadata content management and this means viewers do not have to guess which version of a title to play.
Benefits to media companies include easier catalogue management and less human intervention during content/metadata ingest. Metadata function duplication, where efforts are replicated in the broadcast linear and OTT multiscreen worlds, is eliminated. Metadata consistency means a consistent user experience between platforms and devices, which helps services move beyond ‘TV and multiscreen’ and deliver on the promise of ‘one service, all screens’.
The White Paper also explains how metadata enrichment – where you import data from third-parties to improve the metadata you already have – improves content discovery. Viewers get access to credible, large-sample user ratings from companies like IMDB or Rotten Tomatoes as an additional guide to what they should watch, for example. ‘Cast and crew’ information and synopses can be expanded, giving people more reasons to watch a title or to avoid a title and find something better. Expanded ‘cast and crew’ information provides additional search parameters.
The 3,000 word White Paper provides a clear explanation of the challenges with metadata management today and how these are being solved, and the direct benefits to consumers and media companies when advanced processes are introduced. You can download the White Paper here.