First demo on natural language understanding

I’ve uploaded a new demo on natural language understanding to GitHub along with some background information, see:

 https://github.com/w3c/cogai/blob/master/demos/nlp/README.md <https://github.com/w3c/cogai/blob/master/demos/nlp/README.md>
 https://github.com/w3c/cogai/blob/master/demos/nlp/toh/README.md <https://github.com/w3c/cogai/blob/master/demos/nlp/toh/README.md>
 https://www.w3.org/Data/demos/chunks/nlp/toh/ <https://www.w3.org/Data/demos/chunks/nlp/toh/>

This initial demo was chosen to explore the basic concepts, and future demos will address a larger vocabulary, an expanded grammar, and how statistics, dialogue history and semantic context can contribute to resolving ambiguity in a cognitively plausible manner.

Natural language is typically highly ambiguous, although this is usually unnoticed as we are very good at selecting the intended meaning despite many possible parse trees. To resolve ambiguity in natural language, you need to make sense of the meaning, and to do so concurrently with the syntactic processing. This essentially involves the use of declarative and procedural knowledge in the form of graphs, rules and graph algorithms.

The parser can invoke these via directly invoking graph algorithms, e.g. in an unconscious resolution of lexical ambiguity, or through triggering conscious reasoning in respect to anaphora and prepositional attachment. These processes are asynchronous and run concurrently with syntactic processing.

Suggestions for demo scenarios are welcomed!  These should have well defined language usage and semantics to be practical at this stage of development.

p.s. the initial demo supports speech recognition on Chrome browsers, but Chrome ignores the JSGF grammar passed to it via the Web Speech API and makes frequent mis-recognitions. In future, speech recognition will be much more robust when integrated as part of cognitive agents involving concurrent recognition and semantic processing, so that the context guides recognition.

Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett
W3C Data Activity Lead & W3C champion for the Web of things