IMG_20170606_125123 Discarded exoskeletons of a spiny lobster
IMG_20160604_164221 CYC approach to natural language processing
Dave Schneider, PhD from CYC gives a talk "NLP and Deep Logical Representations" at the Natural Language Processing Day in Austin on June 4, 2016.
This is an example of a text that CYC analyzes and makes all sorts of inferences from it. It talks about Elon Musk making Tesla's patents available as open source for competitors. Based on this text, CYC artificial intelligence software can make business recommendations for a company called Copycat Motors as to what would be a smart thing for them to do when that happens.
To "understand" the text, CYC AI uses something they call Inferentially Productive Representations. I can't say I understood much of it, but very roughly speaking, they take the text and insert into it something that looks like some kind of annotations into it, that tell you something about that concept and its relation to other concepts. That's shown in the next slide. It looks like a Java programmer's fever dream. :-)
This approach to natural language processing is very different from the currently dominating neural networks-based approach.