Empowering Open-Source Generative AI by Integrating the Wikidat...
summary
Practical recipe for plugging the Wikidata knowledge graph into generative models: use a Hugging Face NER/intent step to map user tokens to Q‑items, run a SPARQL query wrapper for live facts from Wikidata or private Wikibase/Wikibase Cloud instances, embed returned triples or text into a vector DB, then feed top matches into a decoder (GPT/LLama) as a RAG prompt. Notes on scale (10–30M entities → billions of strings), periodic hackathons, citation-backed answers, and embedding‑based vandalism detection.
AI-written summary · about
chapters (14)
- 0:00Wikimedia Deutschland intro
- 2:26Open-source community support
- 3:08Wikidata knowledge graph
- 5:16Wikibase Cloud for private graphs
- 7:21SPARQL for Wikidata
- 9:05Toy RAG with Wikidata
- 10:29Live updates and community
- 13:18Wikidata to embedded vectors
- 15:29Global community expansion
- 15:52Wikidata 5M research project
- 18:00Wikidata structure in AI training
- 19:04Future of Wikidata in GenAI
- 21:34Open-source collaboration
- 23:57Q&A on embeddings
related talks
31mWikidata Knowledge Graph to Enable Equitable and Validated Ge...
15mHarnessing the Power of Generative AI for Intelligent Applications
53mGenerative AI: Overview and use cases
37mBuilding Generative AI Applications
19mWisdom-Driven Knowledge Augmented Generation at Scale
24mDataEngBytes 2023 - MEL-T2-03 - Joshua Yu
17mThe keys to success with Generative AI and Vector Databases
27mChallenges of Generative AI