Backward chaining is a system of logic used by artificial intelligence systems. It is designed to solve a problem by working backward from an end goal through a series of rules. This approach can be used by a wide variety of systems, from programs that solve chess games to algorithms used to identify unknown objects. The basis requires robust programming with a set of logical and useful inductive rules the system can use to accurately move through a series of options to arrive at a solution.
In this method, the system is provided with a set of rules by the programmer, who presents it with an end product or goal. The system works backward through the rules to determine how one might arrive at the end goal. In the backward induction used by programs that solve chess games, for example, the computer can take the position of the pieces and move through a series of if-then statements to determine the likely course of movements through the game. A computer can also use backward chaining to explore other possible solutions and branches that could have occurred during the game to change the outcome.
Systems that use backward chaining can have rules that vary in complexity, depending on the kind of work they need to do. A system capable of identifying flowers, for example, may need a large set of branching options to accurately pinpoint the species it is looking at. It could start with a series of statements related to color, move through types of flowers, numbers of petals, foliage, and other characteristics, and determine the identity of a given flower answering questions at each step to determine a final answer. Errors in this process could lead to identification mistakes.
Get startedWikibuy compensates us when you install Wikibuy using the links we provided.
This reasoning system calls upon simple logic. The system examines a fact, determines if it fits with a given product, and takes another step from there. If the fact does not match the available information, it is incorrect, and the backward chaining logic can discard that fact and others that might branch off from it. Facts that do fit allow a program to work with the logic and explore the branching facts to see which fits best. This can work well for a variety of tasks.
Artificial intelligence is not the only entity that can use backward chaining. Researchers who work with primates note that some species appear to utilize this logical method to solve problems. This illustrates a capacity for understanding problems and developing a system to address them.