Machine Learning and OTT Personalization
Learn more about OTT personalization at Streaming Media East.
See complete videos and other highlights from Streaming Media West Connect on Streaming Media's YouTube channel.
Read the complete transcript of this video:
Nadine Krefetz: Where are we with machine learning in terms of how it integrates into personalization?
Randa Minkarah: Well, you absolutely cannot really have it without it. I want to dovetail with what Justin and Jerry here have just said about the complexity of just bringing all of this data together and making it useful. An example on the content owner side is when you think about distribution and realize that there are so many cable systems out there that could be running the exact same content, as Jerry has said, at completely different ways and times. And then there's also the ability to play it back later. When you, as a content owner, come back and try to pull together your numbers on that, you're asking what your audience was from every single source. That's an enormous lift. Inside of some content companies, it takes a day to pull all that together. We literally wrote a program to pull that together. That's the beauty of ML. We can help really bring that data together--even if it's disparate data--and allow you then to start using it. ML is absolutely vital to personalization without the crunch that we've got now with the cloud, With elastic search, with all of the databases that allow you to integrate all kinds of data in near-real-time, you don't have it. And consumers today demand it. They demand that it is up to date is really tailored to them.
Gerry Field: It's a leap from looking at Google Analytics to the idea of getting to those reports a whole lot quicker. To Randa's point, I'm not being as eloquent as she was with it, but it really is the case that we used to talk about big data and managing big data. Well, we're dealing with infinite data at this point. Being able to go through a collection of data and being able to glean any meaning out of it is really a very important thing. A lot of times, we do traditional carriage reports for stations, but we also need to just keep on upping the game on that to really try to get a sense of who's actually watching what for how long in a given space. Again, the idea of melding between traditional broadcast and streaming is really what we're very interested in. We're only gonna see more of that happening. Machine learning is going to be more and more important.
HBO's Sarah Lyons explains how HBO approaches algorithm-driven recommendations and human curation in personalization on the HBO Max OTT channel in this clip from a panel at Streaming Media West Connect 2020.