Unverified50% confidenceFactExact time
Bilibili's recommendation algorithm relies on user behavior sequences to build interest profiles; new accounts initially receive generalized recommendations, and by intensively watching thirty to forty videos in target sections, the account vector can quickly be pulled toward a specific interest cluster.
1
Sources
50%
Confidence
Long-term
Relevance
10/28/2026
First Seen
Sources
Related Entities
Related Claims
Unverified推荐算法本质上是注意力分配引擎,将有限的用户注意力优先分配给已产生高参与度信号(转发、点赞、评论、停留时长)的内容71% similarUnverifiedTikTok/Douyin's recommendation algorithm is based on a multi-layer funnel model, where new videos are first pushed to a small group of seed users (approximately 200–500 people), and initial weights are calculated based on completion rate, like rate, comment rate, share rate, and dwell time.69% similarUnverified抖音、微信公众号等内容平台采用「兴趣推荐+冷启动流量池」的分发模型,新内容先获得种子流量,根据完播率、互动率等指标决定是否进入更大流量池68% similarUnverifiedThe WonderClip homepage recommendation feed performs intelligent recommendation ranking based on user behavior profiles (such as playback duration).68% similarUnverifiedWhen a platform's recommendation algorithm becomes the decisive factor in whether content can be seen, creators are compelled to adjust their topics, duration, thumbnails, and narrative pacing around algorithmic preferences.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1107776API
curl https://kongchang.com/api/v1/knowledge/claims/1107776MCP
get_claim(id=1107776)