Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Musk announces AI ranking and vector retrieval changes to X's timeline

AI summary

Elon Musk said on October 20 that X's home timeline recommendation algorithm is getting two updates: a new AI ranking model intended to predict more accurately how long users spend on a post, and a retrieval-stage change that turns users and posts into vectors to surface content from accounts a user does not follow. IT之家 reported that posts from followed accounts come from a separate system, so the retrieval model determines which strangers' posts can enter the timeline; researcher Rohan Paul was cited saying X already used post completion rate as a predicted behavior signal, making this mainly an optimization of an existing system. Musk also showed a curve of total active user time on the platform, which he said keeps setting records, but per 虎嗅 he did not disclose the coverage or actual effects of either change.

In-depth insight

Because retrieval decides whether un-followed accounts' posts are candidates at all, the change can affect what gets seen and not only how it is ordered, though no public deployment data shows the real impact.

IT之家 10 月 20 日消息,埃隆 · 马斯克今天宣布,X 平台主页“时间线”推荐算法迎来两项更新:引入全新 AI 排序模型,更准确地预测用户停留时长,算法将优先推荐完读率高的贴文。

随后 AI 领域研究人士 Rohan Paul 解读道,本次改进之前,X 平台就将贴文完读率列为排序模型预测的用户行为指标之一,新模型主要是优化已有系统。

同时,第二项改动发生在更靠前的检索阶段。该系统会将用户和每篇贴文转换为向量,再从用户未关注的账号中找出与其最接近的内容。用户关注账号发布的贴文则由另一套独立系统提供,因此该系统决定哪些陌生人的贴文可以进入时间线(IT之家注:Timeline;TL)。

#X#Phoenix recommendation framework

Original · 1 min readRead the original
Group chat

No comments yet. Start the conversation.