Forward reasoning (also known as data-driven reasoning) is used in reasoning systems when rules are run 'forwards' in order to work out further things form the knowledge one has. For example, given "Scrates is a man" and "all men are mortal", one can work out that "Socrates is a man", or ina. numerical example, given "x>y", "y>z", and "z=3" one can work out in one step that that "x>z" and "y>3" and then in a second reasoning step (through towo paths) that "x>3". Forward reasoning is particularly useful when the forwards branching factor is smaller than the backwards branching factor, for example when there is a relatively small numbers of known facts and rules to combine, or when there is a need to precomputer derived knwoledge as a pre-computation step in order to speed up reasoning after deployment.
Used on Chap. 3: pages 39, 40; Chap. 15: page 347; Chap. 18: pages 427, 429, 430; Chap. 21: page 519
Also known as data-driven reasoning, forward chaining, data-driven, forward
See also backward reasoning