Tag: Wikimedia Deutschland

Wikimedia Deutschland announced the Wikidata Embedding Project, a new database that applies vector‑based semantic search to Wikipedia’s structured data. Developed with Jina.AI and DataStax, the system supports the Model Context Protocol and improves retrieval‑augmented generation for large language models. The publicly available database on Toolforge offers richer, multilingual results and will be highlighted in a developer webinar on October 9th. Leggi di più

The Wikimedia Foundation’s knowledge base, Wikidata, is receiving a new AI‑optimized vector database. Built by Wikimedia Deutschland, the project converts millions of Wikidata entries into vector representations using a large language model, making the data easier for AI developers to ingest. Hosted on free infrastructure provided by DataStax and created with Jina AI technology, the backend upgrade aims to level the playing field for smaller AI firms and improve representation of niche topics while leaving the public-facing Wikipedia experience unchanged. Leggi di più