Page MenuHomePhabricator

CJOPP 10,000 page test
Closed, ResolvedPublic

Description

The purpose of this batch test is to test "scale realism," without approaching the final stage (T421607) which contains hundreds of thousands of pages.

Before this test runs:

  • The zhwikisource local RFC regarding whether to implement a new namespace should be concluded.
    • Conclusion: There is no consensus on having a new namespace, thus the decision is to remain the status quo, i.e., no new namespace.
  • The privacy request workflow should be finalized, with a dedicated mailing list.
    • Conclusion: A dedicated mailing list is created. (T423520)

The sample is 10,000 new 判决书 pages (specifically, 10,000 new pages with *判决书 and *事令 title patterns). The date ranges of these judgement documents will be sampled from October 2024, then September 2024, August 2024, etc., till 10,000-page goal is reached.

Event Timeline

Supergrey1 changed the task status from Open to Stalled.Mar 29 2026, 7:42 AM
Supergrey1 updated the task description. (Show Details)
Supergrey1 updated the task description. (Show Details)
Supergrey1 changed the task status from Stalled to In Progress.Apr 19 2026, 10:26 AM
Supergrey1 claimed this task.

Since all blockers are resolved, this task will start today.

Upload has started with October 2024 documents.

From a 1% sampling review, we spotted several minor formatting issues that can be considered "sic" due to that the source is formatted the same way. Still, editors are advised to fix them when they spot them.

From the batch, three privacy issues are found, and then are Revision Deleted:

  • 2 privacy issue under Article 10(1) because quoted evidence preserves an exact doorplate-level address, or exact home addresses remain at street / village-group / house-number level.
  • 1 privacy issue under Article 10(1) because a full unmasked citizen ID number is printed for a natural person.

Rates are relatively low (0.03%) but privacy is indeed something that cannot be taken lightly. The privacy request workflow is expected to respond to such issues that any readers can report.