Basic Knowledge Representation and Reasoning
Declarative Approach
The declarative approach to building agents includes telling the agent what it needs to know. From this the agent uses reasoning to deduce relevant consequences. It requires:
- A knowledge base / ontology
- Which contains fact and general knowledge about a domain in some formal language.
- A reasoning engine
- That produces relevant consequences of the knowledge base.
Example
We want to know what day it is from the following crude knowledge base:
- If I have an AI lecture today, then it is Tuesday or Friday.
- It is not Tuesday.
- I have an AI lecture today or I have no class today.
- If I have no class today, then I am sad.
- I am not sad.
From this knowledge base we can infer that the day is: Friday.
A machine would use reasoning algorithms to solve particular problems in the knowledge base.
Languages for KR&R
To store knowledge in a knowledge base and so reasoning. you have to represent the knowledge in a formal language that ca be processed by machines.
Rule-based languages and propositional logic are KR&R languages.