![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
AO3 Exchanges: Finding Migrated Tags
Aug. 8th, 2017 07:55 pmThere is a known issue in AO3 tag sets where some tags (usually, but not always, AO3 canonicals) will "wander" or "migrate" from the fandom they were initially placed in to other fandoms, probably related to the first, but possibly not. If these migrating tags aren't fixed (by replacing them with more specific disambiguated versions associated with the right fandom and only the right fandom), problems can arise during the sign up stage of exchanges.
While doing some of the admin for Round 2 of Fandom Growth Exchange, I hit upon a spreadsheet-based method to search for these migrating tags. As far as I'm aware, exchanges are currently relying on people spotting these issues manually, so I thought it would be worth writing up my method in case others want to use it too. (Given that tag sets are publically available, it doesn't even need to be mods who do it; third parties who just want to be helpful can too.) Fair warning: it's very kludgey, somewhat time-consuming (possibly a couple of hours or so, for a large tag set, though the time taken can be reduced by working collaboratively once the spreadsheet has been set up), and throws up a lot of false positives (weeding these out is what makes it time-consuming). However, I believe that it should (allowing for human error creeping in at the weeding out stage), find all migrating tags in a tag set.
( Step-by-step method below the cut )
I've just done this to the current state of the![[community profile]](https://www.dreamwidth.org/img/silk/identity/community.png)
femslashex tag set (just shy of 500 fandoms), and it took about an hour, although I was also writing the instructions up at the time. The false positive rate (i.e. the proportion of tags occurring more than once which weren't migrating) was just under 95%. (On Fandom Growth, we had somewhere around 85%, but I think that tag set is structured differently and has more potential for migration: femslashex uses a lot of umbrella fandoms, and by the time I did this exercise they had already been doing some clean up.)
While doing some of the admin for Round 2 of Fandom Growth Exchange, I hit upon a spreadsheet-based method to search for these migrating tags. As far as I'm aware, exchanges are currently relying on people spotting these issues manually, so I thought it would be worth writing up my method in case others want to use it too. (Given that tag sets are publically available, it doesn't even need to be mods who do it; third parties who just want to be helpful can too.) Fair warning: it's very kludgey, somewhat time-consuming (possibly a couple of hours or so, for a large tag set, though the time taken can be reduced by working collaboratively once the spreadsheet has been set up), and throws up a lot of false positives (weeding these out is what makes it time-consuming). However, I believe that it should (allowing for human error creeping in at the weeding out stage), find all migrating tags in a tag set.
( Step-by-step method below the cut )
I've just done this to the current state of the
femslashex tag set (just shy of 500 fandoms), and it took about an hour, although I was also writing the instructions up at the time. The false positive rate (i.e. the proportion of tags occurring more than once which weren't migrating) was just under 95%. (On Fandom Growth, we had somewhere around 85%, but I think that tag set is structured differently and has more potential for migration: femslashex uses a lot of umbrella fandoms, and by the time I did this exercise they had already been doing some clean up.)
