Google is using machine learning to help it to “bring together TV content with the seamless experience of digital” on every screen and surface.
In a post on Google’s blog, Rany Ng, Google’s director of product management for video advertising, said the company’s TV Content Explorer is capable of automatically classifying and making recommendations for how content should be organised across dimensions like show, genre, trending and so on.
“Over the years, we’ve rebuilt our video platform from the ground up — we knew that TV was a very different experience from the web and we knew that broadcasters had different challenges, infrastructure, distribution partners and content from web publishers,” Ng wrote.
“With TV coming to digital, we put our stake in the future of building for a better user experience — one that was connected, always on, and on-demand.”
“Leveraging Google’s machine learning expertise and smart heuristics, TV Content Explorer creates and automatically organises an intuitive catalog of shows and clips,” explains Ng.
“We analyse millions of signals from video content feeds, automatically applying classifiers and making recommendations for how content should be organised across dimensions like show, genre, trending, dayparts, etc.”
“The Explorer will also proactively surface deeper insights into audiences and monetisation opportunities via insight cards,” wrote Ng. She said the new developments are “scratching the surface” of what’s possible.
Google has also launched new forecasting and pacing models, currently in beta, to better predict inventory volumes across multiple devices for TV content.
The new models include a lookback window of 18 months, consideration for organic growth over time, audience seasonality and one-off anomaly corrections for unpredictable events like breaking news.