Help:New filters for edit review/Quality and Intent Filters

From Linux Web Expert

File:PD-icon.svg Note: When you edit this page, you agree to release your contribution under the CC0. See Public Domain Help Pages for more info. File:PD-icon.svg

New filters for edit review introduces two filter groups—Contribution Quality and User Intent—that work differently from other edit-review filters. The filters in these groups offer probabilistic predictions about, respectively, whether or not edits are likely to contain problems and whether the users who made them were acting in good faith.

Additionally, the language-agnostic revert risk model, enabled in 2024, provides a prediction about how likely an edit is to require reverting.

Knowing a bit about how these unique tools work will help you use them more effectively.

These filters are only available on certain wikis.

Based on machine learning

The predictions that make the Quality and Intent filters possible are calculated by ORES, a machine learning program trained on a large set of edits previously scored by human editors. Machine learning is a powerful technology that lets machines replicate some limited aspects of human judgement.

The Quality and Intent filters are available only on wikis where the “damaging” and “good faith” ORES “models” are supported. The ORES “damaging” model powers Quality predictions, while its “good-faith” model powers Intent.

Enabling ORES requires volunteers to score edits on the relevant wiki. This page explains the process and how you can get it started on your wiki.

The language-agnostic revert risk model supports all language Wikipedias and does not require manual training by volunteers.

Choosing the right tool

Looking at the Quality, Intent, and Revert Risk filters, you may notice something different about them. Unlike filters in other groups, the various options don’t target different edit properties. Instead, many of them target the same property, but offer different levels of accuracy.

Why would anyone choose to use a tool that's less accurate? Because such accuracy can come at a cost.

Increase prediction probability (higher ‘precision’)

File:RC-quality-filters-diagram.png
This conceptual diagram illustrates how the Quality filters relate to one another on many wikis (performance varies). As you can see, the ⧼ores-rcfilters-damaging-verylikelybad-label⧽ filter captures results composed almost entirely of problem edits (high precision). But it captures only a small portion of all problem edits (low recall). Notice how everything in ⧼ores-rcfilters-damaging-verylikelybad-label⧽ (and ⧼ores-rcfilters-damaging-likelybad-label⧽) is also included in the broader ⧼ores-rcfilters-damaging-maybebad-label⧽, which provides high recall but low precision (because it returns a high percentage of problem-free edits). You may be surprised to see that ⧼ores-rcfilters-damaging-maybebad-label⧽ overlaps with ⧼ores-rcfilters-damaging-likelygood-label⧽. Both filters cover the indeterminate zone between problem and problem-free edits in order to catch more of their targets (broader recall). For space reasons, the diagram doesn't accurately reflect scale.

The more “accurate” filters on the menu return a higher percentage of correct versus incorrect predictions and, consequently, fewer false positives. (In the lingo of pattern recognition, these filters have a higher “precision”.) They achieve this accuracy by being narrower, stricter. When searching, they set a higher bar for probability. The downside of this is that they return a smaller percentage of their target.

Example: The ⧼ores-rcfilters-damaging-verylikelybad-label⧽ filter is the most accurate of the Quality filters. Performance varies from wiki to wiki, but on English Wikipedia its predictions are right more than 90% of the time. The tradeoff is that this filter finds only about 10% of all the problem edits in a given set —because it passes over problems that are harder to detect. The problems this filter finds will often include obvious vandalism.

Find more of your target (higher ‘recall’)

If your priority is finding all or most of your target, then you’ll want a broader, less accurate filter. These find more of what they’re looking for by setting the bar for probability lower. The tradeoff here is that they return more false positives. (In technical parlance, these filters have higher “recall”, defined as the percentage of the stuff you’re looking for that your query actually finds.)

Example: The ⧼ores-rcfilters-damaging-maybebad-label⧽ filter is the broadest Quality filter. Performance varies on different wikis, but on English Wikipedia it catches about 82% of problem edits. On the downside, this filter is right only about 15% of the time.
If 15% doesn’t sound very helpful, consider that problem edits actually occur at a rate of fewer than 5 in 100—or 5%. So 15% is a 3x boost over random. And of course, patrollers don’t sample randomly; they’re skilled at using various tools and clues to increase their hit rates. Combined with those techniques, ⧼ores-rcfilters-damaging-maybebad-label⧽ provides a significant edge.

(As noted above, ORES performs differently on different wikis, which means that some are less subject to the tradeoffs just discussed than others. On Polish Wikipedia, for example, the ⧼ores-rcfilters-damaging-likelybad-label⧽ filter captures 91% of problem edits, compared to 34% with the corresponding filter on English Wikipedia. Because of this, Polish Wikipedia does not need—or have—a broader ⧼ores-rcfilters-damaging-maybebad-label⧽ filter.)

Get the best of both worlds (with highlighting)

File:New filters for edit review - Intent find-the-bad.png
You can get the best of both worlds by filtering broadly but highlighting using more accurate functions. Here, the user casts a wide net for damage by checking to use the broadest Quality filter, ⧼ores-rcfilters-damaging-maybebad-label⧽. At the same time, she identifies the worst or most obvious problems by highlighting (but not filtering with) ⧼ores-rcfilters-damaging-likelybad-label⧽, ⧼ores-rcfilters-damaging-verylikelybad-label⧽ and ⧼ores-rcfilters-goodfaith-bad-label⧽.

The filtering system is designed to let users get around the tradeoffs described above. You can do this by filtering broadly while Highlighting the information that matters most.

To use this strategy, it’s helpful to understand that the more accurate filters, like ⧼ores-rcfilters-damaging-verylikelybad-label⧽, return results that are a subset of the less accurate filters, such as ⧼ores-rcfilters-damaging-maybebad-label⧽. In other words, all “Very likely” results are also included in the broader ⧼ores-rcfilters-damaging-maybebad-label⧽. (The diagram above illustrates this concept.)

Example: Find almost all damage while emphasizing the worst/most likely:
  1. With the default settings loaded,
  2. Check the broadest Quality filter, ⧼ores-rcfilters-damaging-maybebad-label⧽.
  3. At the same time, highlight —without checking the filter boxes— ⧼ores-rcfilters-damaging-likelybad-label⧽, in yellow, and ⧼ores-rcfilters-damaging-verylikelybad-label⧽, in red.
Because you are using the broadest Quality filter, your results will include most problem edits (high “recall”). But by visually scanning for the yellow, red and orange (i.e., blended red + yellow) bands, you will easily be able to pick out the most likely problem edits and address them first. (Find help on using highlights without filtering.)

Find the good (and reward it)

File:Thank good-faith new users. New Filters for Edit Review.png
This reviewer wants to thank new users who are making positive contributions. The ⧼Ores-rcfilters-damaging-likelygood-label⧽ filter isolates problem-free edits with 99% accuracy. Filtering for Newcomers and Learners limits the search to these two experience levels, while applying a green highlight to Newcomers (only) enables the reviewer to distinguish at a glance between the two levels.

Good faith is easy to find, literally! So are good edits.

The ⧼ores-rcfilters-goodfaith-good-label⧽ filter and the ⧼ores-rcfilters-damaging-likelygood-label⧽ (Quality) filter give you new ways to find and encourage users who are working to improve the wikis. For example, you might use the ⧼ores-rcfilters-damaging-likelygood-label⧽ filter in combination with the Newcomers filter to thank new users for their good work.

Example: Thank good-faith new users
  1. Clear the filters by clicking the Trashcan. Then select the Page edits and Human (not bot) filters.
  2. Check the Quality filter ⧼ores-rcfilters-damaging-likelygood-label⧽.
  3. Check the User Registration and Experience filters Newcomers and Learners (this has the hidden effect of limiting your results to registered users).
  4. Highlight the Newcomers filter, in green.
All edits in your results will be good edits by Newcomers (users with fewer than 10 edits and 4 days of activity) and Learners (users with fewer than 500 edits and 30 days of activity). The green highlight lets you easily distinguish between the two.

Good is everywhere!

The “good” filters mentioned above are both accurate and broad, meaning they aren’t subject to the tradeoffs described in the previous section (they combine high “precision” with high “recall”). These filters are correct about 99% of the time and find well over 90% of their targets. How can they do that?

The happy answer is that the “good” filters perform so well because good is more common than bad. That is, good edits and good faith are much, much more plentiful than their opposites—and therefore easier to find. It may surprise some patrollers to hear this, but on English Wikipedia, for example, one out of every 20 edits has problems, and only about half those problematic edits are intentional vandalism.[1]

Filters list

On wikis where Quality and Intent Filters are deployed, some filters may be missing due to a better quality of predictions. The better ORES performs on a wiki, the fewer filter levels are needed.

⧼ores-rcfilters-damaging-title⧽

⧼ores-rcfilters-damaging-likelygood-label⧽
⧼ores-rcfilters-damaging-likelygood-desc⧽
⧼ores-rcfilters-damaging-maybebad-label⧽
⧼ores-rcfilters-damaging-maybebad-desc⧽
⧼ores-rcfilters-damaging-likelybad-label⧽
⧼ores-rcfilters-damaging-likelybad-desc-high⧽
⧼ores-rcfilters-damaging-likelybad-desc-low⧽
⧼ores-rcfilters-damaging-verylikelybad-label⧽
⧼ores-rcfilters-damaging-verylikelybad-desc⧽

⧼ores-rcfilters-goodfaith-title⧽

⧼ores-rcfilters-goodfaith-good-label⧽
⧼ores-rcfilters-goodfaith-good-desc⧽
⧼ores-rcfilters-goodfaith-maybebad-label⧽
⧼ores-rcfilters-goodfaith-maybebad-desc⧽
⧼ores-rcfilters-goodfaith-bad-label⧽
⧼ores-rcfilters-goodfaith-bad-desc-low⧽

Revert risk

Filter levels TBD. Uses the Language-agnostic revert risk model.


Notes

  1. These figures come from research that went into training the “damaging” and “good faith” ORES models on English Wikipedia. That’s to say, when volunteers scored a large, randomly drawn set of test edits, this is what they found.