About the project
Orange Labs, which handles R&D for France Telecom, reached out to Momentum Design Lab, an HTEC Group Company, to design and develop trailblazing experiences for their new products.
One such product, Video Genome, utilizes machine learning to enable programmatic and producer-based content to co-exist and influence one another—particularly through enabling viewer sentiment tracking.
Viewer insights are gold to anyone who produces content, but there are often limitations in how deep we’re able to go to gauge sentiment. Keen to better understand, explore, and analyze audience reactions, Orange Labs challenged our team to forge the way and create a system for tracking viewer sentiment throughout a piece of content.
Recognizing that human nature is intricate and multifaceted, our team initiated the project by first conducting research into human emotions. What we found was that there are 27 distinct emotions in total, according to a study by UC Berkeley. Each of these 27 emotions carry subtle nuances, with often just a degree of difference between similar ones. So, to mitigate overload of information and ease the burden of choice, we narrowed them down to eight representative emotions.
We then factored in those eight emotions into a pioneering sentiment streaming system. This system enables viewers to share their feelings as they watch. Rather than just being limited to one or two reactions for the overall piece, they can express their emotions at any point of their viewing—for example, when one of their favorite characters makes an appearance or they see something that makes them mad.
The insights gathered from these sentiments are complemented by backend processes that are focused on improving content. Improvements could mean either making adjustments to current content, gathering input for future productions, or tailoring what’s available to each user based on their preferences.
In pioneering a way to keep track of viewer sentiment, our team has unlocked an innovative way to gauge and understand user preferences. By doing so, they’ve laid the groundwork for an enhanced recommendation system that opens up a world of possibilities for streaming platforms, content creators, and, ultimately, audience who consume that content. The solution is already being utilized by a leading streamer to enhance viewer satisfaction.