My first thought was ‘this is just categories - old news in our Notes world’. Closer reading leads me to think not. The article’s concept is, to paraphrase, allowing users to add categories to categories building a trail of relationships that isn’t clear from the raw documents or their initial categorizations.
Neat idea but I’m not sure if A) it can be done in Notes or B) how it would work when the underlying documents might be changed or replaced.
Read the article, review the forum entries (I posted under DoguBlog) and then post ideas and comments about how this might work in a Notes environment.
Very interesting. Of course, much work has been done to model the brain using AI techniques and neural networks.
Neural networks are good up to a point, but don’t tackle the memory/database nature of the brain very well. I thought the blumpy.org article was interesting in that the tags represent both an article AND a relationship, just as a neuron is (part of) the representation of an article - and has relationships to many other neurons (tags).
The problem with building an ‘inference engine’ is that as the number of tags becomes large, the number of possible relationship paths becomes exponentially large. The brain solves this by deliberately NOT attempting to model all possible relationships. Instead we have a (relatively) small number of connections between a neuron and other neurons (some of which are not physically close). In this way the Brain gains great efficiency. A similar model is human relationships - it would be impossible to know everyone on the planet - instead, each of us is supposed to be 6 relationships away from any other person. This is a very efficient way to organise any large network.
Fortunately, this principle makes the problem of modelling a tag network easier than it first appears. I would say that Notes could make a good database structure because of its support for non-structured data.
As for building the inference engine… I would love to if only I had the time.
Interesting but nothing new … not for anyone in the industry who has spent any amount of time with AI, training neural nets or expert systems… would I attempt any of that in Notes? Likely not…but will be thinking about it now…
As for learning how your brain works, this is an interesting place to start: