Project documentation
https://meta.wikimedia.org/wiki/Community_health_initiative/AbuseFilter#Anti-Spoof
High-level goal
Improve the efficiency of other anti-harassment tools (namely AbuseFilter) by improving anti-spoof capabilities.
Prioritization rationale
If our improvements to AbuseFilter are successful, we can improve their accuracy by eliminating spoof workarounds.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | dbarratt | T178704 Enforce that Equivset Test Coverage Remains at 100% | |||
| Resolved | • TBolliger | T166816 Epic ⚡️ : Accuracy improvements to anti-spoof tools across multiple pertinent tools | |||
| Resolved | dmaza | T65242 ccnorm revamp: add a more sensible interface for normalised comparison | |||
| Open | None | T65217 Augment our AntiSpoof normalization data with Unicode/CLDR data | |||
| Resolved | dmaza | T177711 Notify users about ccnorm_contains_any and update abuse filter documentation | |||
| Duplicate | None | T60093 Cyrillic letter Д д (Д д) could match A | |||
| Resolved | dmaza | T153021 Add some more characters to AntiSpoof mappings for usage in TitleBlacklists | |||
| Open | None | T167163 API error response for list=user&usprops=cancreate with similar username has '$1' instead of real error code. | |||
| Resolved | dbarratt | T174195 More extensive unit testing for AntiSpoof | |||
| Resolved | dbarratt | T177667 Get Equivset Test Coverage to 100% | |||
| Resolved | dbarratt | T174197 Split off AntiSpoof equivset generation and string normalization into its own library |
I'm going to mark this epic ticket as done, as the only remaining subtasks are minimal, and the AHT team has no plans to continue working on this in 2018 or 2019.