Why a post appears
A post may appear because its content relates to your interests, matches your language, is fresh, has meaningful interaction or comes from someone you have not interacted with. Multiple signals are combined and repeated posts/authors are reduced. Pausing to read does not prove agreement with the content. The system can be wrong and benefits from your feedback.
Prioritizing unfamiliar authors
Discovery reserves most recommendation opportunities for authors considered unfamiliar based on retained relationship information: following in either direction, conversations, profile visits and post interactions. This goal applies after access and safety checks and may not be met when too little eligible content exists. Unliking or unfollowing does not automatically erase all relationship history used for this definition.
Learning signals and models
Signals include qualified impressions, bounded reading time, likes, comments, hidden posts, not-interested feedback and selected interests. Post text is converted into numerical representations for content matching. A statistical ranking model may be used when validation conditions are met; otherwise the system uses a fallback ranking method. Not every action changes the feed immediately.
Data boundaries
The post recommendation model does not read private messages or call content. A representation of a private post does not grant access to unauthorized viewers. Blocks, post audiences, private accounts, deletions and moderation status still apply. This is a content ordering tool, not a judgment of a person’s character or identity.
Your controls
The current feed uses personalized recommendations. Choose Hide or Not interested from a post menu to provide feedback; use follow, unfollow and block to manage relationships. The current interface does not offer a separate Following tab or a learning-reset screen. To request a reset of learned data or object to recommendation processing, contact contact@loopwith.me as explained on the Data rights page. Resetting does not delete posts, likes or ordinary account relationship history. To remove those, use the relevant controls or submit a data-rights request.
Safety and review
Recommendations do not guarantee every post’s quality, accuracy or suitability. Report violations rather than only selecting Not interested when there is a safety concern. We may change weights and run limited experiments to assess quality without publishing your private data as part of an experiment. Expanding processing to another purpose requires updating the privacy notice.
Contact
Vàng Văn Kim
Tổ 12 Phường Mường Thanh, Điện Biên Phủ, Việt Nam
contact@loopwith.me