Khái niệm cốt lõi
Social media feeds on TikTok exploit user interests in 30%-50% of recommendations, with factors like following accounts and liking videos influencing personalization.
Tóm tắt
The study investigates the impact of personalization on TikTok recommendations. It introduces a framework to analyze user timelines, identifying exploitation vs. exploration. Factors like video watch percentage, early skip rate, fraction liked, and fraction from following are examined for their influence on personalization. Results show significant differences between user groups in terms of these factors.
- Introduction to Social Media Feeds
- Framework for Personalization Analysis
- Data Extraction and Metrics Calculation
- Comparison with Baselines and Bot Traces
- Factors Influencing Personalization on TikTok
Thống kê
Recommendation algorithms for social media feeds often function as black boxes from the perspective of users.
Our results demonstrate that our framework produces intuitive and explainable results.
We find that the algorithm exploits users’ interests in between 30% and 50% of all recommended videos in the first thousand videos of users’ tenure on TikTok.
Trích dẫn
"We introduce a general framework to examine a set of social media feed recommendations for a user as a timeline."
"Our results show that liking and following are the primary drivers of personalization."