Hacker News Reader

Comment by temp | Hacker News Reader

I'd be interested in finding out what happens when I press the thumbs down/up icon on Netflix, because the match ratings that are displayed to me seem to have nothing to do with either my ratings or with my watching history.

View on HN

AFAIK from some conversations I had at a conference last year with a few Netflix engineers who work on this system, it's sort of realtime. Recommendations are build up using Apache Flink, which provides stateful stream compute.

So your click turns into an event on a Kafka topic which is eventually consumed by Flink, joined with whatever other data sources they need and then written out to a store somewhere. IIRC the various machine learning models exist inside their Flink architecture.

https://ci.apache.org/projects/flink/flink-docs-release-1.4/...

It records your like/dislike, but it doesn’t distribute the change to its database immediately. Same for YouTube or so many other services.

The thing is I don't feel like any of the match ratings in the last year and a half since I've been using Netflix have had anything to do with my interests, the ratings I've given out or with my viewing history.

Unless propagation to their database takes two years, I'm really interested in how they actually arrive at match ratings.

I believe largely because a single "like / dislike" is massively inferior to the old traditional Rating system where I am able to Rank content I like...

The recommendation engine seems to be mostly about pushing their original content. Not shocking there.

Reply on news.ycombinator.com