Page MenuHomePhabricator

Research new rules for Underlinked-tagging
Open, LowPublic

Description

In order to avoid tagging https://en.wikipedia.org/w/index.php?title=Louise_Clough&diff=680666170&oldid=646685728 which is only 2 small sentaces, and has 2 links we should research for a new set of rules or something (perhaps make it more restrictive).

Event Timeline

Josve05a raised the priority of this task from to Needs Triage.
Josve05a updated the task description. (Show Details)
Josve05a subscribed.
Josve05a renamed this task from Reachers new rules for Underlinked-tagging to Reasearch new rules for Underlinked-tagging.Sep 12 2015, 2:54 PM
Josve05a set Security to None.

One idea might be to not tag stub-articles with this template if two or more links is present.

Inside the file: autowikibrowser-code\AWB\WikiFunctions\Parse\Tagger.cs (Line no's: ~155-189). Probably just changing the boolean rules around "bool underlinked" to instead factor in total page word count.

Kizule subscribed.

@Reedy - I wanted to create a Pull Request or branch for this for approval/merge. How do I do this in Phabricator? Is there a doc somewhere on the Phabricator developer workflow? I've only used Git and not SVN so I'd need to do a bit of reading on this also.

@Reedy - I wanted to create a Pull Request or branch for this for approval/merge. How do I do this in Phabricator? Is there a doc somewhere on the Phabricator developer workflow? I've only used Git and not SVN so I'd need to do a bit of reading on this also.

The code isn't hosted in Phabricator

You can make a patch against the github repo at https://github.com/reedy/AutoWikiBrowser if that's easier, and then attach a patch to this bug. I'll then commit it to SVN

I added a semi-arbitrary value since word count isn't the same as the variable "length" here - but here's what I added.

bool underlinked = ((wikiLinkCount < 0.0025 * length) && (length > 800));

Here's the pull Request Link that I added:
https://github.com/reedy/AutoWikiBrowser/pull/3

In T193836 AWB over-tagging stubs with Underlinked template, @kaldari indicated that it appeared that the issue was resolved in 2018.
However, AWB tagged the [[Aydan Hojjatova]] stub with {{underlinked}} in January 2021.
At [[User talk:GoingBatty#Underlinked tag]], user Rogermx stated "My objection is to giving any bot the power to add an Underlinked tag to a stub based on some algorithm. Many stubs simply do not have more terms that fit the Wikipedia standard for linking. A bot has no way of understanding that. Wikipedia used to have a 20,000 underlinked article backlog. It was down to about 2400 a few days ago thanks to the work of many people, now it is close to 4,000 again. My guess is that about 30 to 40 percent of these articles have been tagged inaccurately by a bot."

Would it be reasonable to add a new rule that AWB general fixes would not add {{underlinked}} to any article tagged with a stub template?

Dinoguy1000 renamed this task from Reasearch new rules for Underlinked-tagging to Research new rules for Underlinked-tagging.Jan 22 2021, 6:02 AM