[Prev][Next][Index]

CSLI late announcement

Below are the titles and abstracts for the TINLunch and Seminar to be
held this week. 

			    CSLI TINLUNCH
   Reading: "Babe Ruth Homered his Way into the Hearts of America"
		by Ray Jackendoff, Brandeis University
		    Discussion led by Annie Zaenen
			  noon, February 25
			     Ventura Hall

This paper is concerned with the mapping between syntactic structure
and semantic/conceptual structure. When the one doesn't reflect the
other in a direct way, one can either complicate the syntactic
structure (e.g., by assuming a deep structure that would reflect the
conceptual structure more directly) or one can complicate the
correspondence rules. Jackendoff starts from his own specific
assumptions about the conceptual structure (elaborated in his book
Semantics and Cognition (MIT press 1983) and his paper "The status of
thematic roles in linguistic theory" (LI,1987)) and discusses one case
in which the syntax/semantics mapping is not direct; the one
exemplified in sentences like `Babe Ruth homered his way into the
hearts of America.' He concludes that a syntactic solution to the
problem is not appealing but that one has either to claim that one has
a kind of idiom or that the correspondence rules have to be
complicated.  The issue addressed arises of course in all theories
trying to spell out the syntax semantics mapping; the assumptions made
here are different in their specifics from those that most of us would
make but are stated in a notation that is rather close to a attribute
value representation and they argue for a `surfacey' syntax , at least
in this case, so i hope they are sufficiently close to inspire people
to think about their own approaches to this and similar problems


			     CSLI SEMINAR
		       Implementing a BDI Agent
			   Robert C. Moore
			  2:15, February 25
			     Redwood G-19

The BDI (Belief-Desire-Intention) model of rational agency is a
familiar one around CSLI, having been the focus of the RATAG project
for about three years.  As part of the ICA (Intelligent, Communicating
Agents) project, we are attempting to do a complete implementation of
an agent based on the BDI model.  As always, implementation forces us
to confront issues that we had previously overlooked.  This talk will
focus on a number of those issues including:

a formal semantics for desire that can be used to motivate action;

extending the notion of dependency-directed belief revision ("truth
maintenance") to include the dependency of intentions on desires and
beliefs and the dependency of beliefs on intentions;

combining inference and planning by treating intentions as "assumable"
propositions that one encounters in trying to infer that one's beliefs
will be satisfied.