> 3. Use probabilities to guide the inference. Big expanse there, but things can definitely be done this way. I'm not 86% sure, but I suspect this [1] coolness has been done with scraping + Bayesian analysis, i.e. probabilistic stats. (The categories are machine-generated, btw) If anyone does know exactly how this was done please whisper - off list if need be ;-) [1] http://vivisimo.com/ Cheers, Danny. (I like neural nets, me)Received on Wednesday, 24 April 2002 20:39:41 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:53 GMT