Spotlight on Tag Wrangling

The system of tagging that AO3 uses is categorized as a ‘curated folksonomy’, a term which here means ‘a free-for-all tagging system supervised to some degree by a band of merry curators’. Though there are a few other ‘curated folksonomies’ out there, the one used by AO3 is unique in many ways, making the tag-wrangling system an innovation in archival organization and cataloging. (Want to read more on curated folksonomies? Check out (Curated Folksonomies and Tag Folksonomy and Tag Synonyms).

At AO3 the Tag Wrangling Committee organizes the curation which is carried out by approximately 160 tag wranglers. The following is a conversation with part of the committee — Co-Chairs Emilie Karr & Alison Watson, and staffers Jenn Calaelen, Franzeska Dickson, and Sam Johnsson. For the sake of brevity and readability, the replies have been compiled as a group response, and wording has been altered to make answers more understandable to a general audience.

Q: What were some of the first things you learned in order to get up to speed as a Tag Wrangling staffer?

A common challenge is learning to utilize the various tools that are available to OTW staffers and volunteers. This includes a site known as Basecamp, and the chat system, Campfire. Newer staffers are still finding out how things work there and the various things it can do. Because volunteers coming into the OTW have varying skill levels, not all of them have much experience with social media or wiki browsing and editing. Tag Wranglers use an internal wiki system for documenting rules about tagging, as well as recording committee minutes and other material pertinent to their work. This means that learning the tag wrangling guidelines that already exist is a separate learning curve from using the sites and tools themselves, and different members of the committee tend to become more familiar with some parts of tagging guidelines than others.

Q: What have been the big challenges for the committee this term?

One aspect has to do with helping tag wranglers manage their experience as volunteers. The staff need to make sure people are able to keep up with the workload they’ve taken on. It’s easy for a wrangler to assign themselves too many fandoms, and then because of surges in AO3 activity or technical issues, vacations or other personal reasons, it may turn out they haven’t wrangled a fandom in a few months. If one of their fandoms is big or active, that becomes a problem because it’s difficult to catch up when they are able to wrangle again. And the staff also needs to be concerned with wrangler morale, trying to keep both wranglers and staff from getting too discouraged by internal and external criticism of their work.

Another aspect has to do with taking on work most tag wranglers either can’t do or feel intimidated about doing. This includes matters such as cross-fandom content, or things that don’t belong in any specific fandom. That’s not explicitly a committee-level thing, but wranglers are usually more comfortable sticking to things that have clear guidelines and that don’t have a risk of trespassing on someone else’s fandom. Sorting out these types of tags is always an ongoing process and the committee is looking forward to better wrangling tools that will improve mass wrangling (many fandoms at once).

The last issue relates to problems on the Archive that also affect readers and posters. If users are seeing 502 error messages, that means that wranglers have been seeing them for a week or more because the wrangling interface used by wranglers gets lower priority than the AO3’s user interface. So if the Archive’s suffering from technical issues, wrangling may come to a complete halt. But since users may be able to continue posting their work even when wranglers are at a standstill, then new tags that are being created are going unwrangled for longer periods of time. As mentioned before, this is a particular problem with high priority fandoms, those getting a lot of additional works and readers. The AO3 filters being down has meant that wrangled tags are even more important to finding things, while also meaning that a lot of wrangling work is less useful because users aren’t able to use tags to their full extent. An increase in 502s usually coincides with lots of new traffic, meaning more new users who don’t yet know how to find things on the Archive, and new works with new tags that need to be wrangled. Lastly everything in the list of feature requests that Tag Wrangling has made to the Accesibility, Development & Technology Committtee (which runs the archive) gets pushed farther down the priority list because dealing with a technical crisis is always of the first importance.

Q: Do all of you work on the same projects or do you work on different tasks?

We have a rotation for meeting minutes, other tasks we tend to self-assign. We’ve brainstormed lists of things that need to get taken care of. We check in (on Basecamp) before meetings to talk about what work we’ve been doing. Most projects tend to have one or two people focusing on them, but everyone else provides advice and support. In a given week there are various routine tasks such as working on the internal wrangling newsletter; hanging out in chat in case any wranglers come by with questions; training new wranglers when we have an influx; checking on the status of various fandoms to see what may be falling behind; putting things onto our wiki pages, after we’ve made decisions, or we’ve clarified the guidelines already there; discussing support requests; and preparing posts for the wrangler mailing list (usually discussions involving new/changed guidelines, in order to get feedback from wranglers).

Q: How would you describe the Tag Wrangling Committee’s role in creating and maintaining guidelines for tag wrangling?

We’re the maintainers and the advisory committee. Some changes come because of user requests we receive through the Support form on the AO3. User mandate tends to be the rule unless it’s a contested issue among groups of users. A lot of guideline questions are raised by wranglers and we listen to the full list’s feedback. But the staff is needed to coordinate the majority rule, and then it makes the decision in cases where there is no strong opinion emerging. The Tag Wrangling staff are then responsible for writing the wording to try and reflect the decision correctly and then announce all new guideline decisions to the wranglers, to make sure everyone’s informed. Not everyone sees or remembers the messages so we poke wranglers who aren’t following them. This is generally done by leaving comments on the tags themselves, which is something we can see behind the scenes but is currently invisible to the public.

Q: What has been involved in preparing these tag wrangling guidelines for public viewing?

Immense gratitude for Sam Johnsson who is coordinating this work. We have been using the guidelines going public as an excuse to give them a full overview. There’s been three kind of “translations” involved in this preparation. The first is purely technical, translating wiki code (which was used for its posting to the internal wiki) to HTML (for posting it on a public website). The second is some rewriting for greater clarity. A document that faces internally to people who use it regularly reads differently than a document intended for people who are sitting outside, learning about these issues for the first time, and who aren’t clear about the steps that wranglers would then take once reading them.

But the biggest problem has been catching jargon. This is not just because people tend to develop terminology that works as shorthand, but also because over time various different terms have come into use for the same thing on the AO3. For example the tags that users can filter on are commonly called canonicals by tag wranglers but are also labeled ‘common’, a possible holdover from the term coders put into text back when the AO3 was still in alpha stage and it simply never got changed. Similarly, “freeforms”, another common tag wrangler term, was changed to “additional tags” in the forms that users see because initially — in closed beta — on the new works page they were just called ‘tags’, and a lot of users were inclined to put anything in there, from fandom to character tags. The term “additional” never fully transitioned to behind-the-scenes use, possibly because that doesn’t work as well as a noun! But the term “additional tags” makes more sense to users than “freeforms” since the meaning isn’t as clear as “character” or “relationship” tags. The tag wranglers are still struggling with the best way to express the the filterable/canonical/common dilemma since “canonical” is easily confused with a series canon meaning.

You’ll be hearing more from the Tag Wrangling Committee on various tag-related issues in the coming months as our filters return and the guideline publishing is completed. In the meantime, if you have questions about tags or the committee, you can always send a question to our Support team, who’ll pass it on to the Wranglers. The Tag Wrangling Committee also has a Twitter account at ao3_wranglers for all sorts of tag-related discussion.

Archive Of Our Own Newsletter – September 2012

Hello Archive creators, users, and lovers! The days have gone by quickly, haven’t they? We’ve got a short–but sweet–newsletter for you this month.

Cool stuff on the archive!

Releases 0.9.0 and 0.9.1 were deployed — check out what’s new and what’s been fixed! Tag hierarchies are now visible. We’re also able to support podfic embeds again.

What’s up in the world of tags?

With the new public tag display pages, users on the Archive can now get a wranglers-eye view of tags. So if you’re curious, you can check out all the Marvel comics fandoms or all the different adaptations of Hana Yori Dango on the Archive, see how many different characters Stiles Stilinski has been paired with, and hop between a variety of Alternate Universes!

Support issues

We’re starting to see, due to the dedicated and diligent work of our Coding volunteers, a large number of bugs being squashed. Even so, we still have a number of bugs that we know about, along with common workarounds, at the Archive’s Known Issues page. If you want to check to see if something has been reported, that’s a good place to start. (There’s even a link at the top of the Support form.)

We’re excited about the upcoming changes and want to extend a hearty greeting to all our incoming holiday exchanges, especially the number of exchanges who are choosing to run entirely on the site for the first time! If you as a user or a mod have any questions, drop us a line!

As a general warning to users posting HTML, both Firefox and Chrome seem to have resumed their trick of converting straight (“) double quotes into angled ‘smart’ quotes, which our HTML Sanitizer does not like. If you’re having issues embedding multimedia or adjusting CSS, that would be the first place we recommend checking.

AD&T Committee business of note

We’re also working hard on bringing back tag filtering and we hope to have it back soon!. Please take a look at this post for more information on tag filtering if you haven’t already. We’re also discontinuing support for IE6&7. See this post for details and how you can help us with this transition. Release 0.9.0 and 0.9.1 went well and we’re very pleased with how they are working out. And finally, we continue to work on an updated AO3 roadmap and its development is going well!

Support Committee business of note

We’ve made notable steps over the last year to maintain a sustainable system for ticket responses, and, having figured out what works, are starting to collect documented guidelines for the fateful day in the distant future of complete staff turnover.

Tag Wrangling Committee business of note

We’re really excited that everyone can now see some of the insides of the wrangulator, an alpha feature which you may have read about in A New Look at AO3 Tags. We’re currently working hard on converting our wrangling guidelines for posting to the archive FAQs section, starting here.

Apologies

We’ve been getting reports from people who have been missing account activation emails — please do not hesitate to let us know if this is the case with you. We’re working on the issue and hope to have it resolved soon but in the meantime please contact Support if you are missing an activation email. The Archive also experienced about 40 minutes of downtime earlier this month when we installed new batteries in old machines and another 40 minutes during a deploy. We apologize if this inconvenienced you in any way.

Questions? Comments?

We welcome feedback from users! If you have questions or comments, feel free to leave them in the comments of the latest news post, or send in a Support request (if you’re reporting a bug, please send that to Support, as they’re super efficient – comments on our news posts sometimes get overlooked).

A New Look at AO3 Tags

One feature that’s been asked for repeatedly since the AO3 launched is a way to see how tags are structured on the Archive. This is now possible! Although this new option doesn’t show everything that goes on behind the scenes when tag wrangling, it does provide more information to users as to what tags are in use and how the tags are interlinked behind the scenes.

We should note right up front: this is a very alpha interface. In fact, all that’s been done is that we’ve changed the accessibility of the pages, and truncated the display of longer tag lists to save the servers. We do have plans for improving them, but we thought it best to get the pages out there and see what information you want!

We should also note that, in preparation for this release and the guidelines release, we’ve discovered a number of wrangling terms that are used inconsistently or confusingly. The Tag Wrangling volunteers are currently discussing new terminology, which we hope will be clearer.

How to Access the Tag Display Page

There are two ways to access these pages: through navigation, and through direct links.

If you’re browsing the site, you probably already know you can click on a tag to see the works that use that tag or its synonyms. You’ll see early in the page “1-X of Y Works found in Tag Name”. Select the Tag Name to access the Tag Display page for that tag.

If you know the name of the tag, you can also enter http://archiveofourown.org/tags/TAGNAME directly into the address bar. (Note: for relationships, replace any / in the address with *s* for the link to work, and for friendship tags replace any & in the address with *a*.) Tag Search also links to the Tag Display pages.

The tags page

Near the beginning of the page (in the top right when using the default visual skin), will be two buttons that let you see all of the works marked with the tag you’re viewing. Both creators and users of the works may have chosen that tag — the creators when the works were uploaded, and the users when they decided to bookmark those works.

The tags page is divided into several sections. In most sections, if multiple tags are listed, they’re automatically sorted into alphabetical order. Please note that these sections at present will only display the first 300 tags, in order to prevent unwieldy server loads. In the meantime you can use the Tag Search to find a particular tag. We plan to improve this display in later versions of this feature, so eventually you will be able to see all the tags under any tag.

A Sample Entry

A good example to see the Tag Display page in action is the Being Human (UK) fandom tag. Accessing the page, you’ll find that:

  • the tag is a Fandom tag, and it has been marked Common, so it will pop up in the auto-complete;
  • the tag is for a TV Show;
  • the tag has a synonym;
  • the tag has a metatag, indicating that it’s a distinct part of a larger group (in this case, there are other versions of the series);
  • the tag contains a number of Character, Relationship, and Freeform tags.

What does any of that mean?

All the tags on the Archive can be in one of three states — common (canonical), merged (synonym), or unfilterable. Common tags (also known as “canonical” tags) can be filtered on and appear in the auto-complete. Merged tags (also known as “synonyms”) are connected to a single common tag and works/bookmarks tagged with it will appear in the common tag’s filter. Unfilterable tags cannot be used in filters but can still be searched and will still bring up lists of works. Here is an example:

If you click on the merged tag Aido Hanabusa, you can see that this Character tag has been merged into a different tag due to spelling differences. If you then click on the Aidou Hanabusa tag, you will see that it is the common tag. It has various mergers and it is also connected to both broader and narrower tags.

Any tag can be merged if it has a common meaning with another tag or tags. This is true whether it is a Fandom, Character, Relationship or Additional tag. However, not all tags get merged. Some remain unfilterable both because they have no shared meaning with other tags, and yet they are so rarely used that they are not likely to be searched on by other users. Some may be only temporarily unfilterable, until a tag wrangler has had time to review them and mark them as common or merge them with another tag.

Here is an explanation of the other sorts of tags you’ll see on the tag pages.

Parent tags

Each user-created tag has one or more Parent tags. These are broad terms which may contain many subgroups of tags that fit a certain theme.

  • For example, Fandoms will have their Media type(s) listed as their Parent. All television show fandoms will have “TV Shows” as a parent tag.
  • Characters, Relationships, and Additional (or Freeform) Tags will have one or more Fandoms listed as their Parent.

Tags with the Same Meaning

These tags are “synonyms” of another tag, which have all been merged into the common tag. There are various reasons why tags are merged, such as spelling variations, fanon names when canon only gives part of the name, or just that there are many different ways to describe the same thing. When tags are merged they all get pooled together for better filter results.

Metatags and Subtags

Metatags are common tags that can include one or more subtags that are subsets of the metatag. Metatags are created for a number of reasons — the most common reasons are:

  • Fandoms that include different media productions or different media formats under the same name or within the same universe
  • Ambiguous versions of more specific tags (such as all characters named “Mary”)

So if you click on Star Trek: The Next Generation you will see it is a subtag of the larger Star Trek universe metatag.

You can also see its subtags, in this case the movies associated with the series, and you can see that those subtags can have subtags of their own.

Child tags

Like Parent tags, above, each tag can have Child tags. Different types of tags can have different Child tags:

Media tags
These tags, like “Books & Literature” or “Movies” can have Fandom tags.
Fandom tags
These tags can have as Child tags: Character tags from that fandom, Relationship tags that involve one or more characters from that fandom, and Additional tags (called “Freeforms”, here) that are specific to that fandom.
Character tags
These can have Relationship tags that involve specific characters, if set by the wranglers.

Other tags

Besides all of these user-created tags, some tags on the Archive are standardized and cannot be wrangled, though they are still tags and their tag pages are visible. Users are probably familiar with Archive-created tag types such as “Warning”, “Rating”, and “Category” (the latter is for Gen, F/F, F/M, M/M, Multi, and Other). For example, clicking on the Choose Not To Use Archive Warnings tag will tell you “This tag belongs to the Archive Warning Category. It’s a common tag. You can use it to filter works and to filter bookmarks.”

Guidelines are coming

In addition to visible tag structures, the Tag Wrangling Committee is working on making the guidelines that tag wranglers work with available for public viewing. An initial FAQ post about this process is now available. It provides more detail about both terminology and some general concepts.

Questions

For those who have questions about tags and what they’re seeing, you can always send a question to our Support team, who’ll pass it on to the Wranglers. The Tag Wrangling Committee also has a Twitter account at ao3_wranglers for all sorts of tag-related discussion.