Juifeng: scraped and collected content

Hurricane (飓风) is Baidu’s algorithm against scraped and collected content. Version 2.0 defines four distinct offences, and the distinctions matter because three of them catch sites that believe they are publishing legitimately.

The four types

• Obvious scraping traces. Large volumes of content lifted from other sites or public accounts, unintegrated: broken layout, missing functions, unclickable links carried over from the source, poor readability. The tell is not the copying itself but the debris it leaves.
• Content stitching. Several articles spliced together with no coherent logic — text that does not read continuously and answers nothing. Baidu names the cause directly: collection editors and automated assembly tools.
• Site-wide collection. Most of the site is collected; the site itself has little or no production capability.
• Cross-field collection. Gathering content unrelated to the site’s own subject to capture traffic — Baidu’s example is an education site publishing food articles. This is the one worth underlining: it reduces domain focus, and Baidu states plainly that lower focus damages the site’s search performance.

What is explicitly permitted

Republishing with the source stated. Baidu says news and policy content republished with attribution is not treated as collection. The offence is unattributed volume without added value, not quotation.

The penalty, which is unusually explicit

Search display is restricted in proportion to severity, and:

• First offence: restrictions lift three months after the problem is corrected.
• Second offence: restrictions are not released.

There is no second chance in the ordinary sense. A site that cleans up, recovers, and relapses is finished in Baidu search — which makes the editorial policy behind republishing a permanent commitment rather than a campaign.

What a publisher should take from this

Attribute what you republish, keep to your own subject area, and never run automated assembly. Domain focus is a ranking asset here in a way it is not in Google, and the cheapest way to lose it is a traffic-chasing article about something the site has no business discussing.