One of the very best recommendation engines I've encountered is the "Discover Weekly" playlist from Spotify. It's helped me reconsider my relationship to music which I basically thought was dead since I had hit a rut on exploring new artists.<p>There's an interesting presentation of how it's created on SlideShare<p><a href="http://www.slideshare.net/MrChrisJohnson/from-idea-to-execution-spotifys-discover-weekly/8-Discover_Weekly_Started_in_2006" rel="nofollow">http://www.slideshare.net/MrChrisJohnson/from-idea-to-execut...</a>
I'm skeptic about recommendation feature. My reaction to them, in these rare occasions when not get ignored all together, in all sites ranges from 'huh?' to 'wtf?'. Specifically Youtube gives geographic location too much weight -- just because I live in a certain country, it doesn't I'm interesting in all these trending local shit-pop music or stupid 'fun' videos.
Author here - happy to answer questions about the techniques in the paper. We're super excited to finally share this work externally. Feedback about YouTube recommendations in general also welcome.
I noticed that YouTube's recommendations had suddenly gotten better! I wondered if they were using a new statistical approach, or had just started really optimizing at all because the old recommendations were extremely naive. I'm actually a little disappointed to find out that it might just be another deep learning thing. (Yes, it works, but I feel like you learn a little less about problem structure when what you read is mostly "we threw a generic function approximator and 30,000 hours of GPU time at it".)
During the last few months, YouTube has consistently recommended me videos I wasn't interested in (to put it in polite terms), in spite of the fact by now Google knows enough about me to answer quite reliably what I'm likely to be interested in. The only explanation that I can find is that their need to show me specific videos (what do they call it nowadays? “sponsored content”?) prevails over other considerations.
I don't feel like anyone has gotten recommendations right, even though one seemingly obvious approach has not been tried by anyone: allow ratings of favorite works across all media: movies, tv shows, books, music, radio programs, youtube videos. Make a very easy, efficient UI to add ratings. This way you will avoid superficial matches: if I just watched an excellent steampunk cartoon, let's offer a zillion of throwaway crap steampunk. It's not the steampunk part that I liked, it's that it was amazingly done.<p>If I was a huge fan of books, movies, music, youtube picks of another user, it may be there is a deeper connection of the kind of quality we are both looking for, and so his or her recommendations would be highly relevant.
Recommendation systems are a really interesting topic to study/engineer on. I think there's a lot of unexplored/undiscovered techniques still remaining.
Go through the comments of this video <a href="https://www.youtube.com/watch?v=tGe4uWEvwe8" rel="nofollow">https://www.youtube.com/watch?v=tGe4uWEvwe8</a><p>People were literally bizarred by youtube, saying they were there by recommendation. (I have this video in the recommendations also...)
I hate that one day I happened to click one video and watched it, then youtube starts to recommend videos on the same topic day after day, even if I marked them as not interested, they still show up time to time...
I'd be more interested if YouTube recommendations took into account users I've blocked.<p>Sometimes YouTube recommends me videos with clickbaity gross thumbnails or from YouTubers I dislike or have no desire to watch but there is nothing I can do to to stop it recommending these to me, why can't I just go to these users profiles and block them and have them removed from my YT experience?<p>Block just seems to stop people from messaging you, not from you being shown their videos by an algorithm.
I use YT mostly for new music, interesting documentaries and the occasional fun. Beside that there are one-off searches for random topics. Regarding the latter recommendation won't help, because it's not fast enough to tell me which aspect I'm missing. Regarding the former three I'd love to know whether there are users who liked the same videos. So please, recommend users not videos and let me do the rest.
Your recommender seems to be trying to predict (from logs) which video the user will watch next/soon, and how much watch time it will lead to.<p>If you used to have a bad recommendation system, and then you switch over to this system, then it will still be trained with data generated by users who saw the old recommendations, leading it to have a bias towards the same bad predictions.<p>Is there any way around that?
Like others here I am also disappointed by the youtube recommended videos. So I was investigating building a better recommender myself. I was actually searching for how the youtube recommender works yesterday but could only find the 2010 paper. Now I am starting to believe that it is not the recommender that is the problem. It is that youtube consists of 99% low quality videos.
Is this the source of all those Recommendations that I look at Ben and Holly cartoons in the middle of the day when my daughter is at school? Or more Italian daytime television when I've just watched a video my wife never would? In short, is this the reason why there's never anything interesting for me when I go to the frontpage of youtube?!?!
You seem to be doing all the right things in this paper, yet user sentiment seems to still be negative. Do you think that's because maximizing watch time and impressing users are conflicting goals, or are there perfect recommendations out there which both impress users and maximize watch time, yet they haven't yet been found?