This project is my Master's Thesis. It is a folksonomy implementation using an ART-1 neural network. The implementation of this system was a bit of a challenge. The quality of a classification is very good and you see it in action by clicking on the link below. Unfortunately, there are no possibilites to implement it in the production environment due to a very big memory usage. For a big applications that contain a lot of user data, the neural network would need at least 40 GB of RAM in order to work correctly.
You can see an ART-1 based folksonomy in action here: