I would be interested in hearing more about what you are doing. My day job
is as a developer on the Emergent Neural Network Simulation System (
http://grey.colorado.edu/ccnlab/index.php/Main_Page) and I can also provide
insights into feature dimensions from my past experience on automatically
tagging Wikipedia articles with quality (
http://wikimania2007.wikimedia.org/wiki/Proceedings:BM1). I was thinking
that a non-neural network solution, such as as the Core Vector Machine,
might be more appropriate given the size of the training dataset. But NN is
an interesting idea. Maybe we can converse on wiki-research-l?
Cheers,
On Mon, Dec 29, 2008 at 6:13 PM, Soxred93 <soxred93(a)gmail.com> wrote:
See [[User:Crispy1989]]. ClueBot is being rewritten,
so it has an
artificial neural network now. In other words, it has a brain. This
enables it to learn about current vandalism strategies, and start
reverting them without Cobi directly programming in heuristics.
X!