Wikimedia CH/Grant apply/WikiPortraits 2026 Technical Projects
Infodata
[edit]- Name of the project: WikiPortraits 2026 Technical Projects
- Amount requested: 14,000
- Type of grantee: Organization
- Name of the contact: Kevin Payravi & Jennifer 8. Lee
- Contact: kevin
wikiportraits.org, jenny
wikiportraits.org
The problem and the context
[edit]What is the problem you're trying to solve?
[edit]WikiPortraits is a global initiative improving Wikimedia’s coverage of people and cultural events. Since 2024, our 150+ recruited photographers across 35+ countries have uploaded 70,000+ images to Wikimedia Commons, used across 150+ Wikipedia languages and receiving ~400 million monthly views.
Our primary bottleneck is no longer participation, but the lack of scalable infrastructure on Commons. Most of our contributors are experienced photographers new to Wikimedia; they need a clear, seamless upload experience, as well as effective ways to easily explore metrics for their contributions.
Based on our photographers' feedback, we have identified several core gaps:
- Intensive Onboarding: Most of our contributors are new to the Wikimedia ecosystem. Learning categories, structured data, and copyright is a steep curve, slowing onboarding and growth, and requiring training capacity.
- Inefficient Bulk Uploads: The Upload Wizard lacks proper support for high-volume uploads and structured data, leading to poor metadata quality. Existing alternatives are fragmented or lack structured data support. As a result, WikiPortraits photos lack standardization, and ~45% of our images lack "depicts" statements.
- Insufficient Metrics: Current tools fail to provide on-demand reliable usage data at scale. This data is essential for partnerships and credentialing photographers.
These challenges affect the broader ecosystem, from individual contributors to GLAM institutions.
What is your solution to this problem (please explain the context and the solution)?
[edit]This project delivers two solutions: a bulk-upload portal and a mediacount metrics dashboard.
Tool 1: Bulk Upload Portal
[edit]The standard Upload Wizard experience has been described as "clunky" for new users, especially when distinguishing between titles, captions vs descriptions, and structured data "depicts" tags. Photographers expect a streamlined experience, similar to platforms like Flickr. Missing categories or Wikidata items force users to leave the Upload Wizard; an upload experience that requires a user to switch contexts to create new categories or make new Wikidata items (on a completely separate platform!) is significant friction for Commons contributors.
Our proposed Upload Portal would be a new platform for high-volume, event-based photography. Its goal is to standardize event names and descriptions, for better organization and discovery. It will also integrate guided category and Wikidata item creation, keeping users within the upload context while ensuring metadata completeness and consistency.[1] Users would go from potentially having to touch three different "experiences" (Upload Wizard, Wikimedia category creation, Wikidata item creation) to doing it all on one single unified workflow.
This tool will drastically reduce upload time and would be extensible for any bulk contributor on Commons. At the conclusion of this project, our goal is for our photographers to answer "yes" to these questions: "Do you find that your user experience has increased?" and "Do you find that you want to upload more?".
Point person: Vegard Hanssen (https://menneske.no/) is a Norway-based WikiPortraits photographer and software developer. Vegard has professional experience building business-critical web and mobile applications, working across technologies like React, Node.js, Kubernetes, and cloud platforms.
Tool 2: Media Views Tool
[edit]Photo usage and view metrics are vital for campaigns and GLAMs, including WikiPortraits, GLAM bulk-upload partnerships, and "Wiki Loves" upload campaigns. However, existing tools like GLAM Tools break down at scale and lack accurate historical data, as they often rely on article page views instead of actual media view counts. The nature of the data can be wildly inaccurate, such as when a photo of Pope Leo XIV that was on the front page of English Wikipedia for several days but then removed would be undercounted by tens of millions of views.
Our goal is to build a lightweight tool on that leverages the Wikimedia Foundation's Mediacounts dataset to allow any user to query photo or category views. Unlike article-view-based metrics, Mediacounts provides real historical data about how many times each file was actually loaded. WMCH's Cassandra platform already surfaces Mediacounts data for onboarded institutions and partners; our goal is to make similar metrics easily accessible to any Wikimedia Commons user without requiring onboarding or institutional support.[2]
Key functionality includes:
- Ability to query single files or category trees.
- Provide counts, graphs, and lists of most-viewed files.
- Queries are done serverside, so they can run in the background and not rely on the user's browser.
- Queries are saved and sharable with others, so metrics can be easily shared without having to re-run queries.
- Particularly helpful for partnerships, grant reporting, and credentialing where metrics are desired.
- Users will login to their Wikimedia account (via OAuth) to use, since queries are done serverside.
After the grant period, the Media Views Tool will continue to be hosted by WikiPortraits on Wikimedia Toolforge or Wikimedia Cloud VPS. This will not create hosting costs or ongoing maintenance responsibility for WMCH.
Point person: Kevin Payravi (https://kevin.payravi.dev) is a US-based WikiPortraits co-founder and software engineer. Kevin has worked as a professional software developer for 8+ years, and has developed a number of wiki-related tools including View it!, Cite Unseen, commons.gallery, WikiAsteroids, and various tools to help the WikiPortraits team.
Project impact
[edit]We have the following metrics we hope and expect to reach in this initial phase for both projects.
- Upload Portal
- 10+ active beta testers
- 500+ files uploaded through the portal
- "Depicts" usage increased to 80%+ (from ~45% baseline)
- 30 to 50% reduction in average upload + annotation time
- Media Views Tool
- 500+ queries run by WikiPortraits members and other users
- Used in at least 5 credentialing efforts or grant reports
Technical notes
[edit]Both projects will be open source, allowing for community collaboration.
Both projects will be designed to be able to be hosted on minimal infrastructure. Our team already has experience with this, with commons.gallery currently being effectively run on a server that costs less than $5 USD/month (just 512 MB of RAM). For the metrics tool in particular, it will be hosted on Wikimedia Toolforge or Wikimedia Cloud VPS depending on technical need. Both are provided for free by the Wikimedia Foundation for community tools. Other metrics tools, such as GLAM Tools, Pageviews, and the GLAM Wiki Dashboard are also hosted on this infrastructure, as well as intensive querying tools such as Petscan.
WikiPortraits will be responsible for maintenance moving forward into 2027 and beyond. Multiple WikiPortraits team members will be added as maintainers on Toolforge to ensure multiple developers have access to maintain the tool (and avoid "bus factor"). WMCH will not inherit any ongoing responsibility for either tool, as they will be fully maintained by WikiPortraits in the future. We have strong incentive to maintain both tools for our own use, metrics, and reporting (let alone for the rest of the community).
Risk review
[edit]- Scope Creep: We have limited the project to two concrete, prioritized deliverables to ensure on-time completion.
- Adoption Risk: WikiPortraits photographers have demonstrated real-world need for these tools, and will provide an immediate base of 100+ users to validate the tools before a broader rollout.
- Maintenance: These tools are designed to be lightweight, easily sustained by the WikiPortraits team beyond the grant period. Our team has multiple software engineers to ensure we can continue maintenance.
- Key-person dependency: Our team has grown to include several software developers. A separate developer has been assigned for each of the projects above, ensuring they can stay focused on their core deliverable and meet our proposed timeline.
Budget
[edit]The ask for this grant is 14,000 CHF, for the period of May 2026 - December 2026.
10,000 CHF is allocated for the Upload Portal; 4,000 CHF for Media Views tool.
Appendix
[edit]
- [1] Regarding guided category and Wikidata item creation: For Wikidata items, the tool will guide users based on the type of entity being created (e.g., actor, event), prompting for a small set of expected fields. For example, for a person, we'd have occupation, date of birth, etc., with auto-complete integration so users can quickly look up values for fields. This helps ensure consistency with Wikidata norms while keeping the process lightweight. An exploration may also be to pull in fields from external sources and databases (such as IMDb for actors).
- The tool will include safeguards to prevent duplication and low-quality entries. The tool will check for existing categories and Wikidata items using search, providing metadata like description, occupation, and date of birth to allow users to quickly identify if a subject already has a category or item.
- [2] Expanding on the current limitations and our goals: There are a number of tools listed on the GLAM Tools page that Wikimedians use to calculate and track media metrics. While useful, they don't fully achieve what we aim to do with our Media Views Tool.
- BaGLAMa 2 tracks monthly pageviews for articles containing files from a given category. This relies on article views as a proxy, and requires continuous tracking over time. There is no reliable way to reconstruct historical usage if tracking was not already in place. Monthly aggregation also misses short-term spikes (e.g. temporary appearances on main pages). BaGLAMa has also had downtime resulting in gaps in historic data.
- The GLAM Wiki Dashboard and WMCH's own Cassandra are excellent platforms for institutions to track their metrics, including media view counts. However, these are designed for onboarded institutions or campaigns and not built for ad hoc queries to analyze arbitrary files or categories without prior setup.
- GLAMorgan enables on-demand pageview queries for Commons categories, but inherits the same limitations as BaGLAMa 2 (incomplete historical accuracy, missed spikes). Results are not persistently shareable (queries must be re-run), and performance degrades significantly for large categories.
- Mediaviews Analysis gives on-demand mediaview stats (which we also aim to do), but only for up to ten individual files at a time.
- Our Media Views Tool fills in several gaps and provides a suite of benefits that other tools do not provide as a whole: (a) ad-hoc queries that can be run by any user (b) support for both individual users and large category trees (c) runs and saves queries serverside, for performance and enabling very large queries (d) allow results to be saved and shared via links without having to rerun queries.