Denny VrandečićThe recording of my @iswc_conf keynote is available online:
Wikipedia and the Semantic Web -- 20 years of co-development, and the future
Drawing on the roots of @wikipedia and the Semantic Web, how they influenced each other during the last two decades, leading to
@wikidata and peeking forward to @wikifunctions and Abstract Wikipedia.
#semanticWeb #abstractWikipedia #knowledge #knowledgeGraphs #knowledgerepresentation #wikipedia #wikidata #wikifunctions
https://videolectures.net/videos/iswc2025_nara_vrandecic_wikipedia_future
Wikipedia and the Semantic Web: Celebrating 20 Years of Co-Development and the Future
In 2005, the very first papers proposed integrating Semantic Web technologies in the nascent Wikipedia ecosystem. This wasn’t just a convergence; it ignited two decades of mutual inspiration and benefit. From this crucible, the work in semantic wikis drew inspiration. Semantic MediaWiki, particularly, which found global adoption at Google, Microsoft, NASA, and beyond. Wikipedia became the bedrock for pioneering knowledge graphs, including DBpedia, Freebase, and Yago. These pivotal experiences directly fueled the development of Knowledge Graph, a term that has since found ubiquitous adoption, and, critically, Wikidata, a project that has become an indispensable, living component of Wikipedia itself. With over half a million global contributors, Wikidata stands as the world’s most-edited wiki, powering one of, if not the, most widely-used public SPARQL endpoint. Its software, Wikibase, has spawned a federation of knowledge graphs, serving diverse domains from museums to language preservation. Furthermore, Wikidata’s evolution into lexicographic data (inspired by ontologies such as OntoLex and Lemon) laid the groundwork for projects such as Wikifunctions and Abstract Wikipedia, a vision first unveiled right here at ISWC 2018 and now an official Wikimedia Foundation project. This takes us to the present and future: Abstract Wikipedia collaboratively confronts the inherent expressivity gap in knowledge graphs, while foundational role in training and the current use of language models can not be overstated. This creates a tantalising confluence of large language models and knowledge graphs, hinting at profound opportunities - and critical challenges -for Wikipedia, the Web, and beyond. As this rich history promises many more years of co-development and mutual inspiration, we will conclude with a forward-looking sketch of open research questions and exciting upcoming opportunities.
screwlisp
Harald Klinke
Harald Klinke
FIZ KDAI Research Group
bLogic.ink
