Grok explained the newly open-sourced X algorithm in 10 clear and easy-to-understand points:
1) When you open the "For You" feed on X, it first collects your personal information, such as the people you follow and the posts you have liked, replied to, or shared in the past—this helps tailor the content you see.
2) Next, it gathers potential posts from two main sources: posts from people you already follow (referred to as in-network posts) and newly discovered posts from across X (referred to as out-of-network posts) that may interest you.
3) In-network posts come from a fast system that stores the latest updates from the accounts you follow in memory, allowing for quick retrieval without delays.
4) For out-of-network posts, it uses smart AI to compare your interests (based on your history) with all posts on X, selecting posts that appear similar or relevant through a matching process.
5) Once it has these candidate posts, it adds extra details for each post, such as the text, images, videos, author names, and whether they are verified or require payment to access.
6) It then cleans up the list by removing duplicates, old posts, content from your own posts, posts from blocked or muted accounts, posts containing disliked keywords, or content you have already seen.
7) Now, a powerful AI model (based on Grok) looks at your past interactions and predicts the likelihood of you liking, replying to, retweeting, clicking, or even disliking each post—scoring each post accordingly.
8) It combines these predictions into a final score for each post, boosting positive actions (likes) and reducing negative actions (blocks), while adjusting to mix diversity so that no single source dominates your feed.
9) Posts are ranked according to these scores, and the top-ranking posts are selected to create your feed, ensuring a balanced mix of familiar and new content.
10) Finally, it performs a last check, removing any deleted, spammy, violent, or inappropriate posts, and then presents you with a ranked feed for you to scroll through.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
