Bit by Bit to Free Knowledge – how Wikimedia Deutschland is helping to shape the digital future

In a digital world where large tech companies are increasingly taking control of data and algorithms, a free and reliable knowledge infrastructure is becoming ever more important. By developing Wikidata further, we want to ensure that structured and verified data is available to everyone – independent of systems owned by individual companies. High-quality data should therefore be directly available and accessible to open source development teams, researchers and non-profit projects without having to rely on platforms with non-transparent algorithms. Our goal is to strengthen digital self-determination, counteract disinformation and create an open knowledge ecosystem that belongs to everyone.

Wikidata is one of the largest open knowledge sources in the world: high-quality, verified and freely accessible data on millions of terms and concepts is available – continuously updated by a global community of over 13,000 dedicated volunteers. This huge open knowledge graph is already used in numerous projects and records billions of data queries every month – and now Wikidata’s potential is being expanded even further.

In 2024, we launched a new project together with DataStax, a leading provider of AI database management, and Jina AI, a Berlin-based expert in intelligent search technologies. The goal is to make it easier for open source developers to access Wikidata data and thereby promote innovative, non-profit solutions in the department of AI applications. This is because many volunteers in the open source scene do not have the financial or technical resources to access large data sets for their projects – and this is exactly where the new project comes in.

The key is the so-called technology for vector embedding: DataStax stores the structured Wikidata data in a special AI-assisted database that enables a semantic search. While traditional structured data requires exact queries, semantic search enables interaction with the data in natural language – the way humans speak and write. This type of processing is crucial as generative AI models are trained on natural language.

The project is currently in a test phase. Together with DataStax and Jina AI, various technical approaches are being tested in several languages – including Arabic, German and English. What is particularly exciting is that the vector database can be connected to so-called RAG (Retrieval-Augmented Generation) systems. This gives generative AI models direct access to the latest Wikidata data – leading to more precise answers and fewer hallucinations, i.e. incorrect or misleading results generated by AI models. In addition, generative AI models are thus made to incorporate marginalized knowledge sources more precisely.

The first prototype could be completed by the end of 2025. The future of Free Knowledge is evolving – and we are right in the middle of it!

Wikidata is a huge collection of knowledge, but how do computers access this information? There is a programming interface for this, known by the acronym API. It works like a door: if you have the right keys, you can get knowledge from Wikidata or add new information. This makes it easier for websites, apps and even smart devices to access Wikidata’s free data. In November 2024, we released a new version of the interface: REST API version 1. It is standardized, more clearly structured and easier to use than the previous version.

To help even more people use Wikidata, there is now also a portal for developers. Here, everyone can find simple explanations and step-by-step instructions to better understand Wikidata and use it for their own applications.

In 2024, we continued to promote open data and digital participation worldwide. Three key projects show how Wikidata and the principle of Linked Open Data have given rise to new partnerships and international exchange.

One important example is our contribution to a better networking of e-government systems. Efficient data exchange between authorities is essential – but incompatible systems and inconsistent standards often make cooperation difficult. To solve this problem, the European Commission presented the “Interoperable Europe” project in 2022. The goal is to better network administrations and make data management more efficient across countries and authorities. A central component is the Interoperable Europe Act, which creates digital solutions and clear rules for data exchange. At SEMIC 2024 in Brussels – a European Commission conference on better digital cooperation between public authorities – we showed how the Linked Open Data platform Wikidata can help to link administrative data in a meaningful way and thus facilitate digital cooperation in Europe.

We were also part of the 2024 International Digital Dialogues organized by the German Federal Ministry for Digital and Transport and the Gesellschaft für Internationale Zusammenarbeit (GIZ). Our goal was to discuss the potential of open data. It is a key to a more equitable digital future, makes it possible to create sustainable information ecosystems and promotes transparency and access to knowledge – especially in regions where digital infrastructures are still being developed. At the digital conference, we were able to demonstrate how Wikidata and Linked Open Data contribute as open data structures to strengthening sustainable information networks worldwide. As a strong international partner, GIZ in particular can use and further expand the reach and impact of Wikidata. Close cooperation can promote digital infrastructures in developing countries and ensure that open data is transparent and usable for all. Further talks on this are planned with GIZ for 2025.

Our participation in the event “Berlin – Capital of Women Scientists” organized by the Goethe-Institut Tokyo was also particularly important. We emphasized the significance of the representation of female Japanese scientists in Wikidata and showed how open data can contribute to a better visibility of women in science. One highlight of the event was an edit-a-thon. The contributors created Wikipedia articles on Japanese women scientists and actively contributed to reducing the gender gap in science. With the support of experienced Wikipedians, a valuable contribution was made to the digital representation of women – further strengthening the partnership between Berlin and Tokyo at the same time.

Libraries are a publicly accessible place of knowledge – so it is only natural that Wikimedia Deutschland maintains many partnerships with them. Two important key projects took place in 2024.

One of them was the collaboration with AFLIA, the umbrella organization of African library associations, which initiated the IWIPALE project. IWIPALE stands for Integrating Wikimedia Projects into African Libraries’ Ecosystem and aims to integrate Wikimedia projects into the ecosystem of African libraries, thus strengthening access to Free Knowledge. Local libraries play a particularly important role in the preservation and dissemination of regional knowledge – but there is often a lack of digital structures to make African content visible. This is precisely where the project comes in: it supports libraries in using Wikipedia and Wikidata more effectively, thus promoting the visibility of regional content. Wikimedia Deutschland worked closely with AFLIA to implement the project – from the funding application to the Wikimedia Foundation to the development of a training program and an online seminar with the National Library of Nigeria.

One main focus is on revising the so-called Authority Control – a system for linking and validating data. Until now, African libraries have often had to rely on sources outside the continent to prove the authenticity of their content. IWIPALE wants to change this by enabling: African libraries to store their own sources and references directly in Wikidata. This will create a fairer distribution of knowledge that strengthens the African perspective. Additional workshops are already planned for 2025.

The second collaboration took place in October 2024. Together with long-time Wikidata volunteers Epidosis, we conducted an intensive three-day Wikidata on-site workshop for 30 librarians from the Latvian National Library in Riga. The goal of the workshop was to show the participants how to integrate Wikidata into their work. Although the librarians had no experience with Wikidata at the beginning, they quickly learned how to query data, add references and update their catalogs with Wikidata. Within three days, they achieved impressive results: 1,841 added references, 386 edited entries and a total of 3,535 edits.

This event was made possible with the support of Wikimedia Deutschland and funding for Epidosis’ participation as a trainer. The workshop thus provided a valuable opportunity to appreciate the work of a Wikidata community member while fostering a new partnership.

To encourage even more people worldwide to contribute to Wikidata, we joined forces with Wikimedia France in 2024 to establish the first international, permanent online course for Wikidata beginners. Participants learn how Wikidata works and how to add new data using low-threshold and interactive formats. The learning units can be completed flexibly and at your own pace, participants can exchange ideas with others in the course forum or on Telegram and receive a certificate at the end. 58 participants from five continents, including teachers and GLAM experts, successfully completed the course in 2024.

The course was originally developed in French by Wikimedia France. Thanks to the translation and English subtitles for the videos – funded by Wikimedia Deutschland – the course has gained international reach and is looking forward to more participants.