Lecture 7 Cheat Sheet – Capturing & Modelling
Uncertain Information from People
Epistemic Uncertainty
• Definition: Uncertainty due to incomplete knowledge or information.
• Analogy: Not knowing tomorrow’s exam questions because you don’t have full knowledge of
the examiner’s choice.
Aleatoric Uncertainty
• Definition: Uncertainty due to inherent randomness in events.
• Analogy: Rolling a dice — even with complete knowledge, outcome is unpredictable.
Probabilistic Uncertainty
• Definition: Models uncertainty using probabilities (likelihood of events).
• Analogy: Weather forecast: 70% chance of rain.
Possibilistic Uncertainty
• Definition: Models uncertainty using possibility theory (what is feasible/plausible).
• Analogy: Doctor says an illness is 'possibly flu' and 'possibly COVID' without assigning
probabilities.
Ontic Interval
• Definition: Interval uncertainty due to real-world variation.
• Analogy: A machine produces rods 9.8–10.2 cm due to tolerance in manufacturing.
Epistemic Interval
• Definition: Interval uncertainty due to lack of knowledge.
• Analogy: You think the price of a house is between £200k–£250k because you don’t have full
information.
Elicitation
• Definition: Process of obtaining knowledge/opinions from experts or users.
• Analogy: Asking doctors to estimate the survival rate of a rare disease when no data exists.
Linguistic Uncertainty
• Definition: Uncertainty due to vague/imprecise use of language.
• Analogy: One person’s 'tall' may mean 170 cm, another’s may mean 190 cm.
Fuzzy Sets for Human Judgement
• Definition: Fuzzy sets can model vague human terms like 'young,' 'expensive,' 'likely.'
• Analogy: A survey asks people 'Is £500 expensive for a phone?' Answers range across
degrees, forming a fuzzy set.
