User:Pfctdayelise/Categories and galleries

Commons is split in using both categories and galleries to display media. There's little consensus about when one should be preferred over the other, if ever. Both galleries and categories have features that make them preferable in some people's eyes. Commons needs a single, usable system that combines the best of both features.

Why use galleries?Edit

  • In main namespace - search works better ('Go' works!)
  • redirects work
  • more flexibility in annotating and arranging contents
  • allows for selective presentation

But unfortunately,

  • Need to be manually updated, so no guarantee they show all relevant files, or more recent relevant files

Why use categories?Edit

  • 'Automatically updated' - if every file is categorised on upload, there should be little need to do category work (admittedly this is still a dream)
  • category intersect tools in future will be valuable
  • a category link is a much stronger link from content<->meaning, than being linked in a gallery
  • show all relevant items, arranged by filename, show file size and filename

But unfortunately,

  • Translation doesn't work as it should
  • Redirects don't work as they should

There is an ideal size for a category. Because of MediaWiki limitations, it has to be less than 200 items, to allow all the contents to fit on one page.

Categories compared to Web 2.0 'tags'Edit

Many 'Web 2.0' sites like del.icio.us and YouTube (plus many, many more) use 'tags'. Like categories, tags are a way of annotating content.

  • Tags are usually not case sensitive (or there is a practice to only use lowercase tags)
  • Tags are not defined in relation to one another
  • As far as I know there is no way to 'redirect' tags from one to another.


  • Categories are defined in a hierarchy
  • In practice, there is a tendency to use all the tags that apply to an item, whereas for categories there is a tendency to only use the most specific category that applies (although this practice varies from wiki to wiki)
  • Some people use tags as their own personal memos. I have a feeling the whole system works better when everyone is explicitly annotating for the benefit of everyone, which is what MediaWiki categorisation encourages.
  • With tags, sometimes users have their own tags for the same item as someone else, whereas with categories everyone is working on the same categories - one piece of content only has one set of categories.


The way forward - the best of both worldsEdit

Commons needs a single system that utilises the best features of galleries, categories and tags.

Jakob's Law of the Web User Experience: users spend most of their time on other sites, so that's where they form their expectations for how the Web works.

So we need a system that kinda looks like tags. I will refer to it as 'annotation' because it won't be purely tagging or purely categorising.

Separate annotation from contentEdit

Have a separate field for annotations. Editing the page should produce a text box for the page content, and a separate box or whatever for the page annotations.


Make uploading a multi-step processEdit

At the moment uploading is a one-step process. This allows people to upload files and ignore all the requirements we have, apart from the actual file!

Instead we need multiple steps, to force the user's attention to what we're saying.

  • Step 1: Get the file.
  • Step 2: describe the file (+ licensing).
  • Step 3: annotate the file, suggesting annotations based on the user's description. Since the system I propose will still be heirarchical, offer easy ways for users to 'see into' annotations to find the correct most-specific annotation.

Introduce proper aliasing or redirection - an 'Annotation' namespaceEdit

Tagging seems to lack this. Instead of having a rule, all tags should be in plural, you usually have to search wiki and wikis to find all relevant content. Unlike tagging sites, MediaWiki gives us control to arrange how the annotations work in relation to each other.

Perhaps a new namespace, Annotation is needed. If we decide that plurals of common nouns will be the standard canonical annotation, then if I want to change the annotation I will go to Annotation:Mice.

(Note that if I want to see things annotated 'mice', I will just go to the main namespace page 'Mice'.)

So at Annotation:Mice, I'll see something like this:

Annotation:Mice

  • Alternative annotations in English:
    • Mouse, Common mouse, House mouse
  • Annotations in other languages
    • Latin: Mus musculus, ...
    • German: Maus, ...
    • Spanish: Ratón, ...
    • Polish: Mysz, ...

Each 'annotation' may have a canonical form for each language, plus alternative annotations for each language. Based on my user language preference, the Annotation page will call the title the canonical version of my language. So if I have Japanese set as my language, instead of going to Annotation:Mice, I would go to アノテーション:ネズミ ("Annotation":"Mouse"). And then it would show the Japanese listings first, and the other languages including English below.

In this way, all the languages are equal. It's not the case that the English name is the true name and the others are just redirects.

At this Annotation page, users can choose the canonical form for each language, and add or remove alternative annotations. They can also add the parent annotations (for annotations retain the heirarchical structure of categories).


So if I upload a picture of some mice, if I put in the annotation field any of the following: "mice" "mouse" "common mouse" "house mouse", they will all transparently appear at "Mice" (or whatever we have decided to set as the canonical English form of this annotation).

There is an assumption that users annotate in the language that is set in the user preferences. Users could override this by typing for example "ZH:北京" in an annotation instead of "Beijing". This could be useful for mutlilingual users who know the name of something in one language but not another.

Arranging annotated contentEdit

Now we turn to the main namespace page "Mice". (Since it's Commons, and most people are searching for media here, we make the annotated content in the main namespace. But you could have a namespace 'Annotated:' or something similar I suppose.)

The annotated page will offer a drop-down list that allows the user to select how they want to see the content arranged:

  • Arrange by user-created system (current default of galleries - make this default for Annotated pages)
  • Arrange by filename, alphabetically (current default of categories)
  • Arrange by date of upload
  • Arrange by uploading username
  • ...

another separate drop-down list Show X items at a time

  • 50
  • 100
  • 200
  • all

"Arrange by user-created system"Edit

How will this work? It will be like creating a gallery, but you can only use the items with the annotation (ie in the category). Users create appropriate headings and arrange the items in them. All unarranged items, plus all items added to the annotation after the last edit, will appear (or be hidden?) in a final section called 'Unsorted items'.

Possibly take inspiration from Flickr's "create a set" feature.