ACL study finds non-English Wikipedia has major quality and bot-content issues
mdlhx · x · 2026-07-29
ACL paper audits non-English Wikipedia and finds widespread quality issues
A team at ACL 2026 reported that non-English Wikipedia is far noisier than its reputation suggests, especially in low-resource and multilingual settings.
- The researchers filtered the entire non-English Wikipedia corpus using a procedure usually reserved for noisy web text.
- They found systematic problems including script/language contamination, repeated template or placeholder articles, and a large amount of bot-generated content.
- The paper compresses these findings into a 4-level quality ranking for Wikipedia editions.
- In three downstream language-modeling scenarios, training on the filtered data matched or beat raw Wikipedia, with the biggest gains on lower-quality editions.
- The work was recognized with an SAC highlight at ACL 2026.
More from Research
- NSF launches 4-year PhD program as predictions say every enterprise may soon run an AI lab — annbordetsky · 2026-07-29
- New paper says AI-written books are flooding Amazon and crowding out sales — TuhinChakr · 2026-07-29
- SenseTime open-sources SenseNova-Vision, a unified model for detection to 3D reconstruction — socialwithaayan · 2026-07-29
- Schmidhuber says AlphaFold overlooked earlier protein-structure prediction work — SchmidhuberAI · 2026-07-29
- New Paper Uses Inverse RL to Extract Auditable Alignment Rewards from Demos — nagpalchirag · 2026-07-29
- AI Tackles Undeciphered Languages Like Linear A and Etruscan — Ars Technica AI · 2026-07-29