B station recommendation algorithm

Reprinted: https://www.huxiu.com/article/295672.html

Station B weighting algorithm of the recommended full here.

Video is: coins, collectibles, barrage, comment, play amount, the number of points like to share a few, update time.

Column is: coins, collectibles, comments, amount of reading, thumbs up, shares, update time.

  1. Video one more barrage bonus.

  2. Content in 2 years, have accounted for the highest share weights 0.6, this can be understood, most platforms are shared weights infinitely high. Go outside to share content attractive to look at, where are the grandfather.

  3. Playback volume and amount of reading is really his brother, both sides are the lowest weight, but compared to comment on these secondary interactions, low weight can understand.

  4. Comments important than the right to collection, in my mind, this is a bit special. Because I know a lot of weight is greater than the platform is the collection of comments. why? Because, in general, a lot of people will see the content of comment, but only feel particularly good content, will collection.

Favorite high weight, which means pay attention to the quality of content, we pay more attention to feel good dry.

Comments high weight, it means focusing on interactive, able to attract more attention to the content of discussion.

Guess you like

Origin blog.csdn.net/zhezhebie/article/details/89466450