New prototype to share! My main purpose in building it is to inspire you to think more about how open content and open generative AI can be combined to create openly licensed, freely available, interactive learning resources that run privately and locally.
This proof-of-concept combines (1) an OpenStax textbook, (2) a knowledge-graph representation of the book’s contents, and (3) an open weights LLM that runs locally inside your web browser. The locally-running LLM means you can chat forever without having to pay per-token costs, and the knowledge graph means you can browse a visual map of the book that is interlinked with the content itself.
Since a demo is worth a 1000 words, here’s a < 3 minute demo:
This is definitely a prototype and may explode when you try to use it. But you can play around with it at genoer.netlify.app.
If you’re also experimenting with ways to combine generative AI and OER, I’ve just released some tools I hope will make your explorations easier. More people experimenting means we discover useful innovations more quickly!
These tools work with OpenStax textbooks in their source format, “Connexions XML” (CNXML), available in Github. For example, you can find their Introduction to Psychology text in Github.
My first tool, openstax-convert, converts an OpenStax book repo into a single block-segmented book.json file (plus only the images the book actually references), ready to drop into a reader, study tool, retrieval/grounding (RAG) pipeline, or other tool.
The second tool, openstax-graph, takes the output of openstax-convert and derives a typed, adversarially-verified knowledge graph. The graph includes concepts, the real-world entities that illustrate them, the sections where they live, and the relationships between them (is-a, part-of, depends-on, contrasts-with, related-to, illustrates), plus an acyclic prerequisite DAG. The graph is exported as a single graph.json file, ready to use in a wide range of applications.
Have fun, and please share what you build!