Wikimedia Commons is an open access media repository providing visual source material for all Wikimedia projects. While Wikimedia Commons hosts audio, video, animation, digital books, 3D structure files, and even tabular files — a vast majority of the media found on Commons are image files; 94.5% of files are SVG, PNG, and JPEG formats.
In addition to the licensing requirements, files must:
be a media file
be of an allowable free file format
be realistically useful for an educational purpose [2]
Uploading images to Wikimedia Commons is a relatively straightforward process and a multitude of image upload tools are available. There is the standard upload wizard for single image uploads; for bulk uploads of image sets, PattyPan is a widely used tool [3] . Others have opted for sophisticated scripts to load images to the repository from sources like Flickr that periodically check for new images as they become available.
At file upload, users add metadata and assign media to categories. Uploaders may reuse existing categories or create their own. Categories are a product of community-driven folksonomy which, while whimsical, result in reduced search precision and retrieval in comparison to the application of carefully curated controlled vocabularies. Additionally, the lack of metadata requirements at the point of upload has impacted data quality across the repository.
Categories act like library subject headings, but they are not required fields. Therefore, most images have only one assigned category. AI and machine learning should be tactics employed to remedy the sparse metadata. (Image: Vershbow & Fauconnier, 2019)
Metadata quality is a long-standing known issue for the Commons community. Additionally, the inability to translate the metadata into other language editions has been an impediment to image reuse across Wikipedia language editions. To address these challenges, the Wikimedia Foundation received USD 3,015,000 from the Alfred P. Sloan Foundation in 2016 to accelerate the development of Structured Data on Commons in a three-year time frame [4]. The goal is to provide the Commons community with a way to add machine-readable, structured data about the media files using Wikibase and Wikidata as the underlying data repositories, thereby dramatically improving search functionality across the repository.
Wikimedia Commons is not going anywhere. It has a dedicated community and acts as the image repository for all Wikimedia projects. The community is actively working on the backlog of feature requests. The deployment of the Wikibase extension to store image metadata in Wikimedia Commons was a major milestone for the community, funded by a $3,015,000 grant from the Sloan Foundation.
Like many of the Wikimedia Projects, novice users have noted that the secret rules, hacks, and workarounds coupled with a lack of structure can be overwhelming and intimidating, making navigation and search difficult [5]. These barriers to participation have resulted in a relatively small but expert active user community for the project. With outreach efforts around SDC and a new community of SDC data evangelists such as Sandra Fauconnier, the hope is that these former barriers evaporate through iterative feedback between GLAM communities and Wikimedia Technical Teams:
I'd like to encourage all BHL folks to be bold and critical about Wikimedia platforms (and tools), and also express yet unfulfilled, pressing needs and requirements. This would be very valuable feedback for the Wikimedia community, helping us to prioritize improvements to our platforms. (2023, Fauconnier)
To learn more about Wikimedia Commons, you can refer to this helpful guide.
