Feed ranking's goal is to provide people with over a billion personalized experiences. We strive to provide the most compelling content to each person, personalized to them so that they are most likely to see the content that is most interesting to them. Similar to a newspaper, putting the right stories above the fold has always been critical to engaging customers and interesting them in the rest of the paper. In feed ranking, we face a similar challenge, but on a grander scale. Each time a person visits, we need to find the best piece of content out of all the available stories and put it at the top of feed where people are most likely to see it. To accomplish this, we do large-scale machine learning to model each person, figure out which friends, pages and topics they care about and pick the stories each particular person is interested in. In addition to the large-scale machine learning problems we work on, another primary area of research is understanding the value we are creating for people and making sure that our objective function is in alignment with what people want.
Here is the link of VP of engineering.
Machine learning in news feed
- how to get the score, rank of all news
11:20/ 39:49
Model training phase 1 - boosted tree
. Feed has >100K dense features.
.First, prune these to top ~2K
. Training limited by NUM_ROWS * NUM_FEATURES
. Start with 100K features, max rows, keep most important 10K. train 10x rows, repeat
Do this for each feed events, train many forests.
Dispersion
Who is most important people in your life?
Suggesting friends of friends
Actionable Items
I like to figure out how hard it is for me to understand this machine learning algorithm. I think that it is a good idea for me to explore as many topics as I can to prepare onsite interview from Facebook.
No comments:
Post a Comment