How do TikTok recommendations work? (Collaborative Filtering Explained)
HTML-код
- Опубликовано: 24 июл 2024
- Have you ever wondered how recommendation systems on apps like TikTok, RUclips, and Netflix work? The answer is (in part) something called collaborative filtering, which I'll be explaining at a very basic level.
The goal of collaborative filtering is to predict what an individual would enjoy using a dataset of items and a score of how much each individual is attracted to each item. This system is very simple, yet very effective in understanding users' interests, so much so that it's been incorporated beyond the world of social media in e-commerce sites like Amazon.
Thanks for helping out:
Yannie Lam
Karen Li
Cathy Han
Emily Chang
Jason Cao
Credit to all the TikTok creators whose videos I've included.
► Sources:
BOSTRÖM, P., & Filipsson, M. (n.d.). (rep.). Comparison of User Based and Item Based Collaborative Filtering Recommendation Services.
TikTok. (2019, August 16). How TikTok recommends videos #foryou. Newsroom. newsroom.tiktok.com/en-us/how....
► Music in this video:
The Twelve Spies - Silver Maple
Golden Crates - Dusty Decks
Struggles of a Microbe - William Claeson
I Don't Feel Old - spring gang
► Follow me on Tik Tok for more!
www.tiktok.com/@marlonnfuu?la...
Feel free to reach out to me on Instagram @marflon!
#TikTok
00:00 - Intro
00:42 - Recommender Systems and TikTok
01:40 - What is collaborative filtering?
03:26 - Pros and Cons of User-Based and Item-Based Filtering
04:48 - Conclusion - Кино
More tech-related videos to come in the future!
Very informative! Your channel deserves more recognition!
this is super informative and well made
Thanks Kyle! Glad you enjoyed
You did a great job with this!!
Hey thanks Andy!! Looking to make similar stuff in the future
commenting so that RUclips's collaborative filtering algorithm gives me more of this
appreciate that 🙏
liking so that youtube recommends more handsome hunks like this
Great video! So TikTok uses both, item-based and user-based filtering in their recommendation system? :-)