Frames
A structured knowledge representation, "the meaning of the meaning"
Think that a frame is a molecule rather than an atom
Cognative processing w/ frames is bottom up (from natural languang to representation) and top down (from memory to structured representation).
Stereotypes are cognatively efficient, so frames represent stereotypes
Structure of Frames
A frame is a knowledge structure
Ashok ate a frog:
Ate (verb frame)
Subject: Ashok
Object: Frog
Location: Stomach (now)
Time: past
Utensils: ?
Object-Alive: False
Object-is: in-subject
Subject-mood: Happy
Slots (Subject, Object, etc.) & Fillers (Ashok, Frog, etc.)
Slots on a frame can have a pointer to another frame, for instance the Subject
slot above could point to a noun-frame of Ashok. Allows for a discourse level
understanding rather than a sentence level understanding.
Ate (verb)
Subject: -> Ashok
Ashok (noun)
Title: Professor
Location: Atlanta, GA
Properties of Frames
- Frames represent stereotypes of the key word for the frame, like
Ate
- Provides default values
- Takes advantage of inheritance and instances
Animal
num-arms: (default value)
num-legs: (default value)
Human (type of animal) - Class
num-arms: 2
num-legs: 2
job:
name:
Human (type of animal) - Instance
num-arms: 2
num-legs: 2
job: Professor
name: Ashok
- Frames can be used to as a structured representation to store a sentence of input, and can be used to generate a sentence of output.
Representations
Similar to semantic nets, production systems
1) Ate gets added to the working memory 2) The slots and default filler values get loaded from semantic memory into working mem