Wow Schaun, Thank you! You've given me some real food for thought and also suggestions which are very relevant to upcoming decisions I have to make.
I completely agree on not using the chapter/ folder based approach to learning, and that's what the platform is based on, with mastery of one topic potentially opening up many other doors (see image on my reply to you on LinkedIn).
I really like the idea of monitoring which paths students take the most often and feeding that into an algorithm or AI system that helps to adapt the learning path over time. I'm assuming that's also where the suggestion to use tags comes in, as those are a bit more flexible than relationships in a parent/child topic table, and easier for an ML system to manipulate over time?
I had been thinking that the best way to curate the relationships was to have an AI make a first pass at mapping out the relations by methodically digesting a curriculum document. This would then be followed by having human teachers (who use the map as a curriculum map to have suggestions on the order in which to teach their pupils) giving input where they suggest a relationship has been mapped out incorrectly - similar to Wikipedia style editing where consensus is formed over time about the 'true' relationships by a community of experts with an interest in maintaining the node. The aspect of AI being able to observe the paths taken by students is a brilliant addition to this, and I think that to make it work I just to find a good measure of what success looks like for a student, so that I have a good goal for the AI to optimise towards.
Wow Schaun, Thank you! You've given me some real food for thought and also suggestions which are very relevant to upcoming decisions I have to make.
I completely agree on not using the chapter/ folder based approach to learning, and that's what the platform is based on, with mastery of one topic potentially opening up many other doors (see image on my reply to you on LinkedIn).
I really like the idea of monitoring which paths students take the most often and feeding that into an algorithm or AI system that helps to adapt the learning path over time. I'm assuming that's also where the suggestion to use tags comes in, as those are a bit more flexible than relationships in a parent/child topic table, and easier for an ML system to manipulate over time?
I had been thinking that the best way to curate the relationships was to have an AI make a first pass at mapping out the relations by methodically digesting a curriculum document. This would then be followed by having human teachers (who use the map as a curriculum map to have suggestions on the order in which to teach their pupils) giving input where they suggest a relationship has been mapped out incorrectly - similar to Wikipedia style editing where consensus is formed over time about the 'true' relationships by a community of experts with an interest in maintaining the node. The aspect of AI being able to observe the paths taken by students is a brilliant addition to this, and I think that to make it work I just to find a good measure of what success looks like for a student, so that I have a good goal for the AI to optimise towards.