Fact-Checking with Wikidata

​Hands-On Methods Using MCP and Beyond GenAI - Philippe Saadé

​In this hands-on workshop, we’ll explore practical solutions to fact-check with Wikidata. We will start by setting up the Wikidata MCP and giving an LLM the tools needed to explore Wikidata and test claims. Then we will move beyond GenAI and build a small experimental fact-checking system that uses semantic retrieval and a natural language interface classifier for a more controlled and interpretable approach.

​We’ll cover the following steps:

• ​Introducing Wikidata

• ​Setting up the Wikidata MCP

  • ​Fact-check with the MCP and GenAI

• ​Setting up a fact-checking solution beyond GenAI

  • ​Retrieving Wikidata statements with semantic search

  • ​Filtering statements with a reranker LLM model

  • ​Classifying statements for fact-checking with an NLI model

​By the end of this workshop, you’ll be able to understand how to use Wikidata for exploration and claim verification and build a small fact-checking pipeline in a controlled and transparent way.

​

​About the speaker:

​Philippe Saadé is the AI/ML project manager at Wikimedia Deutschland. His current work focuses on making Wikidata accessible to AI application with projects like the Wikidata vector database and the Wikidata Model Context Protocol.

​

​DataTalks.Club is the place to talk about data. Join our Slack community!

​

​This event is sponsored by Wikimedia