This page is currently a draft.
This extension lets you add annotations to files. It currently only supports images but can later be used to add annotations to audio and video files too. It is an improvement on the ImageAnnotator gadget and currently has a similar feature set.
Text and photos and videos are great, but today we don't have the same kind of pervasive hyperlinking in media that makes disappearing down a Wikipedia clicking journey both fun and educational. - From T133526
To achieve this FileAnnotations might benefit by being more structured than ImageAnnotator. Commons has guidelines on what kind of annotations are considered good and what format makes sense given a particular context. Encouraging these guidelines through the user interface, and making it easier to do the right thing, while still keeping the contribution process simple is important.
Users and use cases
Personas are we targeting and what their use cases might be.
- Discover the feature and figure out how to use it .
- Use the annotations for a visual learning experience for her students.
- Get encouraged to add annotations of her own as the contribution doesn't require much time or effort.
- Be able to contribute without learning about too many policies.
- Share his know about modern design without the hassle of creating new articles.
- Get guidance on what annotations work best.
- See his annotations on an article, without having to edit the article. This might be motivate him to get deeper into editing.
- Quickly tag event photos with relevant bands as links.
- Monitor and revert any vandalism on annotations.
- Improve annotations by others, by turning text annotations to links, for example.
What is a good annotation?
The guidelines on Commons on what makes a good annotation are helpful for users, but we need to find guidelines on what is helpful for the entire system. I feel that in some cases a link to Wikidata item would serve better than a Wikipedia article or a commons category. The way we display the link to the Wikidata item can make us reach closer to our goal too (I am thinking something like Popups and not just a plain link).
Given the current use cases we can roughly divide the types into the following:
|Type||Source||Remarks||Do we want it?|
|Category||Commons Category||Will structured meta-data have any affect on this?||Yes|
|Thing||Wikidata or Wikipedia item||While adding we need to show the user only the Wikidata result if there is one. Its easy enough to find out WP entries from there, but not the other way around.
Once a Wikipedia article is added, adding more languages needs to be supported while editing.
|Location||Some geo template?||This is not for the location of the image (that could already be in the EXIF data) but the location of places in it. Might be helpful in skyline type photos.||?|
|Text||This is specifically for transcribing text in images. Not sure if it needs to differ from the Wikitext option.||Yes|
|Wikitext||We could either have this as a separate type, or have the "[[ ]]" icon for each type and let power users edit the Wikitext of pre-decided types too.||Yes, but in what form?|
These types need not be explicitly chosen by the editor but can be inferred and then displayed accordingly. The interface however could however encourage links over text.
How will a reader find out that there are annotations on a particular file?
- Current state (gadget/extension): Text note at the bottom of the file.
- Current state (gadget): Icon on top the thumbnail indicating the file has annotations.
- Flash the annotation rectangles for a short period of time after the page is done loading.
- Flash the annotation rectangles as the image thumbnail scrolls into view in an article.
- Load the annotations and keep showing them till the user doesn't mouse-in-out once.
- Show a list of annotations on the page, separate from the image.
- TextExtracts and PageImages for Wikipedia articles (like Hovercards).
- Title, Image and Description for Wikidata items.
- See more images link, where applicable.
- See on map link, where applicable.
- Read more on Wikipedia.
- Image from commons with a frame.
- If its a text with a link like, "Foot from " then the Wikipedia page hovercard can be embedded below (similar to link embedding in chat applications.
- An embedded map if the annotation has a geo template.
How can we encourage annotating files, both readers and editors?
- Show notification to add annotations based on the category of the image. Eg- people, map etc.
- Are we running any face recognition or focal point recognition services? If yes, show a notification with a rectangle already around the face/focal point.
How should the actual annotating process work?
- Current state: Lets you add wikitext annotations.
- Should we allow for structured annotations? A tab for each annotation type? It could save in wikitext like it already does, and infer the type later again based on the type of the link.
- Don't allow anonymous users to add plain-text annotations?
- Be able to see diffs and watchlist annotation pages (maybe automatically when you watchlist the File page).
- Have links to quickly convert plain wikitext annotations into links.
- Have links to quickly convert Wikipedia or Commons links to Wikidata links.
We could show the different annotation types with a description first and then open a different dialog based on type. I am not to keen on this as it makes the process longer, but it might be easier for beginners to understand what each type is good for.
We could alternatively open a dialog with tabs, similar to the "Insert media" dialog in VE, and give tabs for each type. The Wikitext tab could always be populated, no matter the type, so that power users can edit if they need to.
Flash annotations on file after the page has loaded
Alternatively, the annotations could remain visible till the user doesn't hover and un-hover over the image once
While on an article the annotation boxes could flash as you scroll by images
When images are in certain categories like people, places, we can encourage the reader to add annotations
When an annotation is a Commons category, it could show thumbnails of images from that category and a link to see more
When the annotation is to a Wikipedia article we could show a Hovercards-like interface
When the annotation has a geo template we can directly render a map, that can be linked to OSM or our own maps
When the annotation is to a Wikidata item, we can show the main image and description along with anything we can link to