Case Studies
Here are some case studies of Learning Graphs I have worked on with others. The list is alphabetical and you will find that the maturity of the on-line material varies widely based on the needs and resources of each group I work with. What they all share in common is that they have a course description that was used to drive a generative process to build a learning graph.
In general when we create a website using a
tool like mkdocs, we structure all the content under the docs
directory. Within docs we place our interactive simulations (MicroSim)
under the sims directory by convention. I would suggest
creating a sim called graph-viewer inside the sims directory.
Many of the websites I setup also walk people through the process of
learning to write GenAI prompts to generate the first draft of
a learning graph. If that is the case, the output of these
prompts is located in the `docs/prompts/learning-graph area.
Stages of Learning Graph
There are two paths we have seen. Start with an existing textbook. Textbooks are usually oopyrighted by an author and publisher. So to keep out of legal issues, we don't have any examples of these. But we know these exist. If you do have an existing textbook or group of papers you can run them through an natural-language processing (NLP) tool to extract the entities, concepts and taxonomies. These process are well documented using the GraphRAG architectures.
These cases studies focus on pure knowledge extraction from large-language models.
Steps Generative AI Driven
- Course Description - we usually start with a generic course description as it would appear in a course catalog
- Bloom Taxonomy Refinements - we refine the course description to fit Blooms't taxonomy which starts out with concept definitions and proceeds to creating hands-on skills for creating new content
- Concept Enumeration - create a flat list of about 150 concepts ideally sorted from simple to complex
- Concept Dependencies - for each concept, create a list of what other concepts it depends on
- Concept Taxonomy - for all the concepts, create about ten classifications for all the concepts
- Content Generation - once you have your concepts organized, you can then use this knowledge to create various content
Circuits
This was the course that I worked on with Sharat Batra at the University of Minnesota. He has been an inspiration for me, helping me understand the time-consuming nature of generating circuit diagrams, equations and assessments in his daily teaching work.
Geometry
This is a classic high school geometry course that lends itself to visual thinking.
Graph Algorithms
This website is based on the graph algorithms course I developed for my hands-on graph courses in my consultant practice. I was fortunate to develop graph database related courses that over 3,000 people have taken in the past. The core courses were actually an introduction to graph databases which was an online self-paced course and an instructor-led Hands On Graph course that used TigerGraph. I hope to work with volunteers to get those courses in the public domain sometime in the future.
Learning Graph for Graph Algorithms
Learning Graph
This is a learning graph for creating learning graphs. Yes, it is a bit of a recursive definition, but this learning graph has helped me fill in the gaps of this website and put more focus when a concept needs to be explain clearly for others to participate in the process of building intelligent online textbooks.
MicroPython
This is based on the course that I developed with the help of Jim Tannenbaum. Jim has also used these courses in his volunteer work in schools.
One note about this course, the goal of the course are to teach computational thinking in junior high and high school students. But we have found that the best way to teach these skills is through project-based learning where students create what they are most curious about. So having a rich curriculum full of projects that drive colorful moving lights and drive robots is our best way to get these concepts into the classroom.
Teaching Computational Thinking with MicroPython MicroPython Learning Graph
MicroSims
This course was inspired by the wonderful Val Lockhart and Troy Peterson. Val was the one that coined the term "MicroSim" and showed that by combining generative AI with the p5.js system that instructors could create fantastic interactive simulations. MicroSims are small interactive simulations that run in a browser and are created and modified by generative AI systems for use in classroom instruction.
Signal Processing
This course came about in collaboration with Jarvis Haupt at the University of Minnesota Department of Electrical and Computer Engineering. Jarvis has been instrumental at encouraging others at the University of Minnesota to take a leadership role in the integration of generative AI into their courses.
Systems Thinking
This is one of my favorite courses. It was originally inspired by my friend Arun Batchu. I have taught this course to a wide variety of audiences including variations for teams trying to break down silos within organizations.