Better Knowledge for Better AI
There's a growing consensus that knowledge graphs – which are a kind of artificial knowledge for artificial intelligence – will power the next big steps in AI.?
The challenges facing today's generative AI demonstrate clearly and conclusively that for better AI we need to model not only data (like sequences of words or pixels), but thinking itself as well. To make that happen, we need to recruit the right people and build the right tools to improve the quality and coverage of the structured knowledge that's already available.
These are some of the topics that I talk about in this episode with Larry Swanson on his new Knowledge Graph Insights blog.
Disambiguation Specialist
4 个月Excellent interview Mike Dillinger, PhD and Larry Swanson. The segment starting at 7:45 or so, when Mike explains how 150 million job titles in 23 languages were managed. Here's the money line: "Separating out the concept of Software Engineer from all of its manifestations... We would take the German title for "Softwareentwickler" and connect it to the identifier 69137 which is 'Softeare Engineer - language independent - then the algorithms could work on the IDs." This technique is enormously helpful in all kinds of translation activities and along side the separation of a concept like Role from its many manifestations is another. Very well done, gents!
Fundador y Líder global D.X.O.? (Digital eXperience Operations) | Digital Transformation Advisor - Educator | Acompa?o a mis clientes en la Transformación Operativa - Business Process Management
4 个月So interesting and educational explanation, Mike Dillinger, PhD. As Engineer I have to say a felt a bit frustrated and seriously thinking of doing ome Analytics Lingusitics learning. ?? Primarily interested in using Knowledge Graphs for Personalized Learning, now trying to explore the collaboration between Knowledge Graphs and Business Process Management. I see the rellationship between LLM / Natural Language Processing and Business Process Management clear, but How to fit Knowledge Graphs in the equation?. Any hint on this last topic?. Thanks for the podcast and also Larry Swanson for sharing.
Content Strategy Evangelist | Co-Host of Coffee and Content | Host The Content Wrangler Webinar Series
4 个月I love opening LinkedIn and seeing thoughtful and informative posts like this one from Mike Dillinger, PhD. Thanks, Mike!
Senior Director of Content Platforms and Knowledge Engineering
4 个月Absolutely! In many cases, shops should consider converting and creating content in structured formats that work naturally with structured knowledge models, but unfortunately, engineering teams are not content nor KM experts and everything to them looks like code.
Expert on policy interpretation, rules, concept models, vocabulary, knowledge and data.
4 个月If you model thinking, then you're modeling concepts - not data - as the starting point. I would love to hear more about what you see as the differences, and how they relate to today's graphs under RDF et al.