algorithm Stories July 3, 2015

How likes & other user input work to personalize your Apple Music experience

Apple has talked a lot about how it’s tackling curation better than the other guys by putting a big focus on hand-picked, human curated playlists for Apple Music, but how exactly does a user’s input alter the music the app serves up? The Loop’s Jim Dalrymple spoke directly with Apple to answer that question and put together a guide detailing exactly how likes and other user input work to customize some aspects of Apple Music but not others:

First, let me tell you one of my big problems, or sources of confusion, with likes on streaming services. Let’s say I’m listening to a Metal station and a great song comes on, but I consider it to be Rock. Do I like it? I enjoy the song, but I’m afraid if I like it, more Rock songs will come on the Metal station, diluting it… What if I don’t like it? Will it never show up again, even in Rock? Perhaps I should skip it, but is that equivalent to a “dislike”?

The guide is great if you really want to make the most of the service, but further proof that Apple Music is a complicated mess and not very intuitive for users when it comes to how likes, hearts, and other user input features of the service affect recommendations.

Head over to The Loop for the full guide on how the way you use Apple Music might make for a better, more personalized experience.

algorithm Stories August 15, 2013

Apple bought Matcha because it “found the answer” with its recommendation algorithm

Following reports earlier this week that Apple had acquired video aggregation and discovery service Matcha, TechCrunch claims today that its sources have provided a bit more info on the motivation behind the acquisition. While noting that the purchase price was actually closer to $10-$15 million opposed to the $1 to $1.5 million reported by others, the report says that Apple is after the company’s proprietary recommendation algorithm rather than just its talent:

Nor was it an acqui-hire; this was about the product Matcha built and about the specific recipe for video recommendations it put together via its proprietary algorithm, according to one source close to the matter.Matcha was acquired after testing numerous approaches to generating recommendations, right at the point where it had refined its algorithm such that it saw an explosion in user growth, according to our source. The app did definitely do well on the App Store charts, and was ranked among the top 15 apps in the Entertainment category before it was shut down.’

TechCrunch also adds that Apple found Matcha’s user acquisition and user engagement strategy to be the “best of any other apps competing in that space” and that it had “found the answer” with its recommendation algorithms:

It was Matcha’s user acquisition and user engagement strategy that Apple was interested in, according to one of our sources, since the acquisition happened just after Matcha had completed a round of vigorous A/B testing and had “found the answer” to rapid user growth and time spent in app. Matcha’s pairing algorithms that drove the right content to the right users simply worked best of any other apps competing in that space, the source affirms.

The report also confirms that Apple actually acquired Matcha prior to the company shutting down the service back in May.

The video programming recommendation app, which lets users browse across services like Netflix, iTunes, HBO, Hulu, Amazon Prime and others, didn’t pick up much steam on the App Store after launching in January 2012. While some have speculated that Apple could use the service to help power recommendations for a revamped Apple TV service, today’s report compares the purchase to Apple’s previous acquisition of app recommendation and search service Chomp.

Powered by WordPress.com VIP