The learning group continues and it's still just me. I picked up Toby Seganan's book "Programming Collective Intelligence" from O'Reilly publishing. It's one more programming language to learn (Python) but has some interesting stuff.

So far, it seems to be more about data-mining techniques, but looking ahead, there are sections about neural networks and genetic algorithms. It seems that “Collective Intelligence” means a lot of different things to a lot of different people. Currently there is no unified theory of Collective Intelligence (tho a lot of people are working on that right now). One place where a lot of CI techniques are coming from is Multi-Agent, Distributed Artificial Intelligence, where people are substituted for the “agents”. Done right, this can be very effective. If the agents in the system are trying to decide between different solutions involving time and distance, a computer can handle it with little or no problem. If it is a decision between the esthetics of different solutions, the computer is just about useless, but for people, it’s easy.

I'm gaining new respect for Information Filtering though. It turns out you can get more out of it than what you put into it. You can find ways in which other people are like you and get recommendations based on what people who are like you probably like (User-based filtering). You can find correlations between your stuff and other stuff that you have no experience with (Item-based filtering).

I’m very interested in solutions like this, that can give more than what is put into them. It seems this isn’t a feature in all CI techniques. Some, like Wikis could be called JBOP (borrowing a little from RAID terminology JBOD). A Wikis strength is in its ability to combine the efforts of a lot of people into a seamless whole, but it is still “just a bunch of people”.

I’m realizing that Cmaps isn’t the perfect collaborative learning environment. Concept Maps are useful for representing the relationships between ideas but the more information you put on them, the less useful they become. Wikis are great for organizing a lot of information but don’t give you that immediate sense of the whole of the subject. I like the way Cmaps lets you create links on each concept. Right now I have links to a couple different Wikis. I guess what I’m going to have to do it set up my own unless I find something in Cmaps that would work as well. From what I’m seeing, a lot of collaborative learning environments are a collection of software packages and it is the techniques that make it all work. This is great if you know what you’re doing and not so great if you’re figuring it out as you go along (like me).

I will keep you posted.