Analogy making is widely considered as an essential ingredient of human intelligence. Unfortunately, a clean mathematical model of this process is still missing.
In this talk, we will investigate analogy through diverse viewpoints:
- Analogy as a way to match situations from different contexts.
- Analogy as a way to reason and to 'infer' new conclusions from existing facts.
- Analogy as a way to solve relational equations.
Finally, we will discuss analogy in natural language, and show how classical techniques become things of the past with the emergence of machine learning based techniques.