Conceptual Dependency, Scripts and CYC

(alias: Strong Slot-and-Filler Structures)

"The important point about a representation is that it makes certain information explicit." David Marr

Conceptual Dependency (CD's) is a theory of how to represent the kind of knowledge about events that is contained in natural language sentences. (Schank, 1973, 1975)

• facilitates drawing inferences from statements
• is independent of language

(note: most parsers use semantic nets)

page 278 primitives verbs (events)

page 280 examples and page 279 rules

arguments for goodness - page 282

expectations

- fill in

- answer questions "understand"

expand

causality

assumptions

"Fred believes John"

"John shot the girl with a rifle" with blond hair ...

examples of where first inclination may be incorrect...

want ability to change interpretation

Scripts:

A structure that describes a stereotype sequence of events in a prticular context.

Events described form a causal chain

• prediction (occurance of events not explicit in story) page 286 restaurant ...out of money (pg. 287)
• answer questions (build a coherent story)
• focus attention on unusual events

• fleeting (mentioned briefly and may be mentioned again)-- store pointers

• non-fleeting -- activate fully and attempt to fill in slots
  The headers (its preconditions, preferred locations, its props, roles, events) can serve as indicators that the script should be activated. To cut down how many scripts active, greater than 1 header.

SAM - Script Application Mechanism

Scripts/ Goals/ Plans

MOPs

Memory Organization Packet

a MOP whose strands are exclusively filled by MOPs

Thematic Organization Points: Organized chunks of episodes that have in common some overall theme or goal-oriented point (terrorism, divorce) BORIS

PAM Planning Application Mechanism

CYC - Lenat 1986 - funding (2 million?)

"to encode the large body of knowledge that is so obvious that it is easy to forget to state it explicitly" (book)

"the building of a large knowledge base of real world facts and heuristics and methods for efficiently reasoning over the KB" (abstract)

encyclopedia

arguments: are we ready (do we know how to represent anything well enough that this project will ever be useful) or will representations change so fast that it is never stay the same.

Motivations (page 288)

Ontology (page 292)