I've been trying to learn about knowledge representations.
According to a canonical definition, a knowledge representation is:
- a surrogate for real-world entities
- an ontological commitment about what kinds of entities are relevant
The notion of surrogate is nicely described in the same source:
For any entity (surrogate) in the knowledge representation, we have to be sure we have a good argument for:
- "correspondence": what does this entity in the KR represent in reality?
- "fidelity": how well does it match what it is supposed to represent?
Can someone help me think of a useful heuristic to understand what is meant by a surrogate here? The best example I can think of is a map (which has both correspondence and fidelity )