EFTA02564951

Document Text

From: n behalf of Ben Goertzel Sent: Tuesday, January 22, 2013 1:41 PM To: Jeffrey Epstein Subject: Re: When Hanson makes his robot model of *me•, we can try different hairstyles, and see which one audiences find more compellling ;-) It's my pleasure to introduce you to folks.... Another kind of brilliant AGI guy is Linas Vepstas http://linas.org He has worked for me off and on.... He's unusual in that he is a world-class software engineer (he worked for IBM porting Linux to various supercomputer architectures), a physicist/mathematician (PhD in physics, author of 400+ math/physics Wikipedia pages), and also very smart about AGI.... Linas is not an academic, he's an industry guy.... But he has great ideas about how to implement systems that learn language from experience.... He has been working on a project using OpenCog's MOSES component to predict which Army soldiers will commit suicide based on their psychologist reports (I'm an advisor on that project). The predictive accuracy we've obtained is reasonably high. In his spare time, he is in the midst of re-implementing the link parser http://www.link.cs.cmu.edu/link/ -- historical webpage with online parser http://www.abisource.com/projects/link-grammar/ webpage with current version maintained by Linas which is part of our current OpenCog language processing pipeline.... The goal of his re-implementation is to make the link parser interpret language more like people do... Talking to Linas would g