Problem spaces in natural-language understanding


next up previous contents
Next: How Soar works
Up: How does Soar work? Problem spaces.
Previous: How does Soar work? Problem spaces.
Back: to main list of student notes

Problem spaces in natural-language understanding

Taken from page 461, Proceedings of the 13th Annual Conference of the Cognitive Science Society. (In the Oversize section of the library.)

States are the possible mental representations one can build of sentences etc, plus some notion of what the focus of attention is.

Example of how a sentence might be comprehended:

  1. Initial state: looking at sentence ``John is rich''.

  2. Apply the operator ``Attend to next word''. New state now includes a focus on the word ``rich''.

  3. Apply the operator ``Look up meanings''. New state now includes the following information: ``(1) = has lots of money (for people); (2) = well-supplied with fat, sugar etc (of food)''. It also includes the following partial mental model, from interpreting ``John has'': ``John (person) has (some property)''.

  4. Apply the operator ``Add a meaning to the mental model''. This changes the mental model to ``John (person) has (property) lots of money (of people)''.

  5. Apply the operator ``check semantics''. Are the qualifiers ``(person)'' and ``(of people)'' compatible? In this case, they are.


next up previous contents
Next: How Soar works
Up: How does Soar work? Problem spaces.
Previous: How does Soar work? Problem spaces.
Back: to main list of student notes



Jocelyn Ireson-Paine
Wed Feb 14 23:45:33 GMT 1996