Proposal:Media review

Develop an improved toolset for reviewing large numbers of multimedia files, especially when the bulk of files to be reviewed is of low quality or no educational value.

As a result of the Ford Foundation funded Multimedia Usability Project, Wikimedia Commons is receiving a step-by-step multi-file uploading tool (Extension:UploadWizard). This should significantly simplify basic media contribution to Wikimedia Commons by individuals.

Very large potential remains:

• Smartphones typically come with built-in cameras and GPS chipsets, turning them into networked cameras with geo-coordination capabilities. Millions of smartphone users therefore are potential spontaneous contributors of relevant photographs of buildings, art, people, news events, etc. to Wikimedia Commons.
• Flickr is the single largest repository of Creative Commons licensed photography, with more than 100 million photos under CC licenses overall, and tens of millions under compatible licenses.
• Other educational projects like Encyclopedia of Life, LibriVox, or ConneXions, as well as open access journals like the Public Library of Science journals, release significant amounts of educationally valuable illustrations, spoken books, photographs, and other media content.
• Wikimedia Chapters and other Wikimedia movement players continue to organize photo contests and other media competitions such as Wiki Loves Monuments that require large numbers of media files to be reviewed.

The challenge in importing any of these media files is the process of selecting and correctly annotating them to be useful in the Wikimedia universe. Because some of the potential sources of media files are so full of noise, depending on the origin, some of this review needs to take place prior to upload.

In short, this proposal is to develop:

• a set of plug-ins/apps that interface Wikimedia Commons with various sources of media files (whether it's apps for different smartphones, plug-ins to obtain data through the Flickr API, plug-ins to obtain media contained in PLoS publications, etc.);
• a temporary "stash" for files and metadata that are in the process of being reviewed;
• a set of user interfaces to select files, edit metadata, instantly add categories, etc.
  • optimized for review of "high noise / high discard ratio / no community interaction" media sources such as smartphone uploads
  • optimized for review of "low-medium noise / low-medium discard ratio / high community interaction" media upload directly on Wikimedia Commons

The process diagram below visualizes the basic idea:

The FlickrLickr bot is a historic (now inactive) precedent for a very singularly focused review tool. Users had to apply for an account on the system, and were then assigned "slices" of license-compatible Flickr content to review. At 50 photos per page, users could choose which photos to pass on to Wikimedia Commons, edit the descriptions, and add categories. Some simple helpers force users to enter filenames for files which come with no useful name or title, allow re-using Flickr tags as categories, and allow re-using a single media description page on multiple files.

Some of these UI patterns are likely to be useful in any bulk review tool. Below is a screenshot of the FlickrLickr review interface for a single photo (out of 50 per page):

• What are the most effective user interfaces for reviewing large amounts of media files?
• What information should be asked for at minimum in all scenarios where we're soliciting user uploads (e.g. what level of UI complexity is appropriate for a smartphone uploader)?
• Which repositories are the most likely to provide valuable content?

• Engineering costs (with some ability to distribute costs through a modular plugin architecture), user experience testing and design, additional storage space and system load. Ballpark: To get some basic plugins working and a decent review interface built, all in all likely to be at least a two-quarter project with two-three primary resources, but not requiring full-time dedication of all resources.

Do you have a thought about this proposal? A suggestion? Discuss this proposal by going to Proposal talk:Media review.

1. Eloquence 21:19, 24 November 2010 (UTC) currently still in exploratory mode, but want to investigate this further as a potential major activation area for new participation + more quality content