Researchers Mapped 72 Years of German Parliament to Wikipedia's Fact Bank

Researchers in Germany took every person who ever spoke in the Bundestag — the national parliament — between 1949 and 2021 and matched them to Wikidata, the free shared fact database that sits behind Wikipedia. That database stores structured details like birth date, birthplace, profession, gender, and awards. The result is one long, searchable timeline that pairs what politicians actually said with who they were. Think of it as turning decades of dusty transcripts into a spreadsheet you can sort.

That combination is powerful for researchers and journalists. You can now ask questions like: did women lawmakers speak about different topics than men? How many speakers were born outside Germany? Which professions dominate parliament? The authors also track 'historical legacies' — how the Nazi era and the former East Germany still echo through today's records — plus languages spoken and foreign honors received.

The catch is the data itself. Wikidata is built by volunteers, so how complete a person's entry is depends on how famous or well-documented they are. The paper finds coverage is uneven across gender, profession, and background. Big names get richly detailed pages; backbench MPs, especially women and those from minority groups, often get a few lines or nothing at all. So the dataset reflects not just who spoke, but who the internet bothered to write about.

Almost every AI system today is trained on text and data like this — Wikipedia and its cousins. If the underlying facts are patchy, the AI's answers inherit those blind spots. The lesson for the rest of us: open data is not automatically fair data. It's a useful mirror, but a cracked one — and you should ask who is missing before you trust what it shows.