The program acts like a human that attempts to give intelligent responses according to your input. Programming a computer to appear like a human that understands normal human
language is a very important subject in the field of Artificial Intelligent (AI). A natural-language processing (NLP) technique is usually used to process human language. Basically, the NLP system
parses a sentence, that is, reads sentence word by word and try to ‘figure out’ the meaning of these words. An appropriate response will then be given base on the extracted information.
A classic 'talking computer' is the ELIZA program, written by J. Weizenbaum in 1964. Basically, the program acted like a Gogerian psychoanalyst. Instead of leading the conversations, the psychiatrist takes
a passive role by echoing the patient’s own remarks. In this case, only a minimal amount of computer code is dedicated to parse a sentence.
One of the crucial factor for a computer to appear as an intelligent human being is the sophisticate extent of a NLP technique. Because of the enormous size and complexity of a human language, implementing a foolproof
NLP system is difficult. Obviously, some sort of restrictions will have to impose so that programming task becomes more feasible. By including all vocabulary and all possible ways a sentence may construct is impractical, if not impossible.