Uncertainty in Heuristic Knowledge and Reasoning

Heuristic knowledge and reasoning based on the knowledge is more or less uncertain, while computers behave logically based on rigid principles of their operation. In order for computers to mimic human intelligence using such uncertain heuristic knowledge, they must have a certain model to represent and process the uncertainty. The paper recalls how Artificial Intelligence has dealt with uncertainty; the mainstream was symbolic approaches until late 20th-century, then numerical approaches have become dominant by 21st-cenetuary. The paper summarizes major symbolic and numerical approaches with their characteristics. Then, it introduces our numerical approaches of knowledge acquisition from data and reasoning with the knowledge; a probabilistic approach, a possibilistic approach and an evidential reasoning with knowledge acquired by Rough set theory. The paper concludes with an expectation to possibilistic approaches and combined approaches of evidence theory and rough