ORES review tool

The ORES review tool is the key user-facing feature of the ORES extension, which provides objective revision evaluation services to automatically rate a revision's characteristics: likelihood it is vandalism, degree to which it might be damaging, likelihood of being good faith, likelihood it will be reverted, and overall quality. The review interface integrates the scores generated by the ORES service into MediaWiki's interface. ORES provides automated scoring of revisions in order to aid editors. For example, ORES can predict whether or not an edit is vandalism, as well as the overall quality level of an article. See ORES' documentation for more information about what types of scoring are available.

Patrolling RecentChanges with the review tool enabled
ORES presentation at Wikimania 2016
The default threshold is deliberately set low to capture almost all vandalism cases (so a number of false positives are also likely to occur). This is in contrast to anti-vandalism bots which set the threshold high to capture only the most obvious vandalism cases (and thus have few false positives). If you do not want to see the flag for most edits, you can simply change ORES sensitivity (see below).

Using ORES

If the ORES extension is activated, you can enable the review tool within your user account by looking under the "beta features" section of Special:Preferences. The review tool will augment Special:RecentChanges and Special:Watchlist by highlighting and flagging edits (with a red-colored r) that need review, because the ORES prediction model judges them to be "damaging". You will also be able to filter these lists by selecting the "Hide probably good edits" option. When you select this option, the review tool will hide any edits that ORES judges to be unlikely to be damaging.

If you review an edit and realize it is not vandalism, you can simply mark it as "patrolled", and the highlighting and flag will be removed.

You can change the sensitivity of ORES in your preferences (under the "Recent changes" tab) to "High (flags more edits)" or "Low (flags fewer edits)". You can also choose to make "Hide probably good edits" selected by default.

 
Changes that "need review" and may be damaging are highlighted and flagged in Special:RecentChanges.
 
A "needs review" flag is described in the Special:RecentChanges legend.
 
Configuration settings are available via Special:Preferences.

FAQ

How does ORES detect damaging edits?

ORES uses machine learning strategies to "learn" what damaging edits look like, by reviewing examples created by Wikipedians through Wiki labels.

Why use the term "damaging" instead of "vandalism"?

"Vandalism" is just a subset of what we want to catch when we're doing RC Patrolling. The word "vandalism" implies deliberate malicious intent. However, a patroller's job is to look for damaging edits whether the damage was actually intended or not. Therefore, referring to the edits that the review tool flags as "damaging" is more true to the kind of work the system is designed to support.

Note that the ORES service also provides a model that focuses on the good-faith/bad-faith distinction ("goodfaith"). It'll be easier to take advantage of that when we deploy the next major change to filtering on the RC page for the review tool. See the Including new filter interface in ORES review tool topic under discussion.

See ORES#Edit quality for more information about how "edit quality" is evaluated in ORES.