AI计算广告论文播报 2026-06-12:统一召回与归因绕过 episode artwork

EPISODE · Jun 13, 2026 · 8 MIN

AI计算广告论文播报 2026-06-12:统一召回与归因绕过

from 周六9点半

这一期我们聊两件正在被工业界重新认真对待的事:单一表示能不能取代手工拼装的多阶段召回,以及预测准能不能等同于归因对。两篇论文,分别给出了肯定和否定的答案。本期重点* 统一可编辑的生成式召回(OneRetrieval: Unifying Multi-Branch E-commerce Retrieval with an Editable Generative Model):快手把倒排、向量、协同三路召回收敛到一个生成模型,关键是用可解释属性位+预留空槽,让运营小时级加新词(比如突然爆红的 LABUBU)而不重训模型——这是过去倒排迟迟拆不掉的真正原因。* 预测不等于归因(Forecasting Is Not Attribution: Localizing Decoder Bypass in Graph-Based Neural MMM):作者用一组很狠的对照实验证明,无图、全连接图、真...去小宇宙查看完整单集简介在小宇宙查看该单集文稿

Episode metadata supplied by the publisher feed · Published Jun 13, 2026

Embed this episode

Ready to play

AI计算广告论文播报 2026-06-12:统一召回与归因绕过

0:00 8:18

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

No similar episodes found.

No similar podcasts found.

Frequently Asked Questions

How long is this episode of 周六9点半?

This episode is 8 minutes long.

When was this 周六9点半 episode published?

This episode was published on June 13, 2026.

Can I download this 周六9点半 episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.
URL copied to clipboard!