Knowledge-based Agent
The core component of the knowledge-based agent is the knowledge base (KB) . The knowledge base is a collection of sentences. These sentences are expressed in knowledge representation language and express certain assertions about the world.
When certain sentences are directly given rather than inferred They are called axioms, and those derived from axioms are called theorems.
Two basic operations of the knowledge base: Tell (add new sentences to the knowledge base) and Ask (query what is currently known)
logic
Model checking: by enumerating all models that may be true to verify that α is true when KB is true
Propositional logic
grammar
The grammar of propositional logic defines legal sentences. Atomic sentences are composed of a single proposition. Each proposition represents a proposition that is true or false. Generally, capital letters are used to represent the proposition, such as: P, Q, R
have two propositions The words have fixed meanings: TRUE is the proposition of eternal truth, FALSE is the proposition of eternal false
Compound sentences are composed of simple sentences with brackets and logical connectives
Semantics
model
Implies
Logical equivalence
Proof method
- Model checking
- Logical equivalence
- Inference rules
A sentence is unsatisfiable, that is, this sentence is equivalent to false: (If it is proved that this sentence is unsatisfiable, it is proved that this sentence is false)
A sentence is satisfiable, that is, this sentence is equivalent to true
Model checking
To put it plainly, model checking is an exhaustive method. List the truth table of each variable, and see the true and false of this sentence.
Logical equivalence
Inference rules
Use old sentences to introduce new sentences
Formalized as a search problem
Clause and CNF
Clause: All variables are all disjunctive relations
CNF: main conjunctive paradigm
Resolution principle
Instance
HORN clauses and finite clauses