Knowledge Representation and Expert Systems Flashcards
What is an expert system?
A system that behaves like an expert in some narrow area of expertise.
What are the three main modules of an Expert System?
- Knowledge base
- Inference engineer (shell)
- user interface (shell)
What are features of expert systems?
- Problem solving in area of expertise
- relying heavily on domain knowledge
- interaction with user during and after problem solving
- explanation, ability to explain results to user
What are desirable features of rules in ES?
- Modularity
- Incremental
- Modifiable and transparent
- Represent uncertainty
- Chaining of rules (to create extensive reasoning)
Define modularity in terms of rules of ES:
Each rule define a relatively independent piece of knowledge
Define incrementality in terms of rules of ES:
new rules can be added independently of other rules.
Define modifiable and transparent rule in ES:
Rules are easy to modify or determine what goes wrong
What are some benefit features of an expert system over a machine learning approach?
- Transparent decision making
- Explain ability of reasoning
- Incorporation of domain-specific knowledge
What is the certainty of the event c(P1 AND P2)
If it is AND the probability is:
min(P1, P2)
What is the certainty of the event c(P1 OR P2)
If it is OR the probability is the max between two values:
max(P1, P2)
What is the query format for semantic networks?
?- fact(characteristic(instance name, Output).
e.g
fact(moving_method(kiwi, KimMovingMethod).
What is a frame?
A data structure to represent the components of data on a semantic network
Give a frame for Bird example from lectures (slots: a_kind_of, moving_method, active_at)
FRAME: bird
a_kind_of: animal
moving_method: fly
active_at: daylight
What is machine learning?
Computer systems that are able to learn and adapt without following explicit instructions
What is the slot value frame format in Prolog?
frame_name(Slot, Value)
Examples:
- bird(a_kind_of, animal).
- bird(moving_method, fly).
