Chapter
How YouTube Uses Vector Space Models for Video Recommendations
30:43 - 38:02 (07:19)
YouTube uses vectors in the space of all the videos on the platform to generate diverse recommendations for users based on video views, time watched, likes, comments, and more.
Clips
One way to generate diverse recommendations for YouTube videos is to compare the vector of a video with other vectors of watched videos and finding similar vectors with some degree of diversity.
30:43 - 33:57 (03:13)
Summary
One way to generate diverse recommendations for YouTube videos is to compare the vector of a video with other vectors of watched videos and finding similar vectors with some degree of diversity. However, quality news and journalism rely on a journalism department, despite the possibility of low-quality videos getting high views.
ChapterHow YouTube Uses Vector Space Models for Video Recommendations
EpisodeCristos Goodrow: YouTube Algorithm
PodcastLex Fridman Podcast
The amount of time people spend watching a video and giving feedback through surveys are important metrics to measure the value obtained from it, as it helps in predicting what viewers will like in the future.
33:56 - 36:13 (02:16)
Summary
The amount of time people spend watching a video and giving feedback through surveys are important metrics to measure the value obtained from it, as it helps in predicting what viewers will like in the future.
ChapterHow YouTube Uses Vector Space Models for Video Recommendations
EpisodeCristos Goodrow: YouTube Algorithm
PodcastLex Fridman Podcast
The act of subscribing to a YouTube channel can signal satisfaction and interest in the creator, but some users may misuse the feature by subscribing to many channels they don't watch.
36:13 - 38:02 (01:49)
Summary
The act of subscribing to a YouTube channel can signal satisfaction and interest in the creator, but some users may misuse the feature by subscribing to many channels they don't watch.