This is a project that's been going on for over a decade at a very high RD cost. Their basic epistemology for their conceptual model looks hopelessly flawed. Not only is there ambiguity between instances of concepts and concepts, they even have circular definitions (see the definition of concept on their free browser:
http://opencyc1.cyc.com:3602/cgi-bin/cyccgi/cg?cb-start.)
They admit to extensive contradictions in their system and try to solve it with contextual focusing, but the results seem very unpromising. They also admit that adding new information to the db gets increasingly hard because of the web of contradictions. I think the only hope this project has is for a system with a conceptual model more closely following the human conceptual model scanning their entire db for useful information and creating an entirely new hierarchical structure.