Revert Risk
Estimate the risk that a Wikipedia edit will be reverted, in every language edition of the site.
Revert Risk helps find the Wikipedia edits that need reviewing. Text-based models exist only for some languages and leave out smaller editions; this one can be used on all of them. It answers through a public API, and can also feed bots that assist patrollers and administrators.
And where does AI come in?
An XGBoost model estimates the probability that an edit will be reverted. It looks at article quality and at the contributor: account age, number of edits and of user groups. The model card warns that it may be biased against new users, temporary accounts and IP edits. It recommends a multilingual variant for anonymous edits, where that variant is available.
The AI label
- Model or technique
- XGBoost, on language-independent features.
- Model published
- Yes
- Where it runs
- On Wikimedia's infrastructure (Lift Wing), with a public API.
- Data
- Twelve months of Wikipedia edits (2022), excluding bots.
- Role of people
- The model flags edits for review; its card advises against using it as ground truth to train other models.
- Frugality
- Lighter than its multilingual variant, which needs more processing power and can be slow.
“Not filled in”: MAUB could not check it with the project. Know more? Write to us.
Run this project? Show the “Listed on MAUB” badge
Paste this code on your site: an image and a link to this page. No script, no tracker: MAUB cannot tell who displays it.
<a href="https://maub.subalpine.it/en/projets/revert-risk"><img src="https://maub.subalpine.it/statique/img/badge-recense-en.svg" alt="Listed on MAUB" width="180" height="48"></a>Something wrong or out of date? Write to contact@subalpine.it.
The monthly newsletter
Once a month, what moved on MAUB, in your inbox.