Classification of Guided Expert Systems According to Their Accuracy, Domain Resolution, and Range

Abstract: An “Expert System” is based on “trial and error”. The configurations’ classification is difficult for the human user; however, such categorization may be performed with the assistance of a human experts. The random configurations generated by the computer may be classified and for further computer expert system interrogation by the user.

Imparting knowledge can be performed as follows:

Ÿ Transferring the knowledge — the instructor shows the collection of elements A = {e₁, e₂, …, e_n} to the learning entity, and each value v_i is assigned to the corresponding element e_i (i = 1, 2, …n).

Ÿ Using the knowledge — the learning entity receives an element e,  for consideration, where, and assigns to e its corresponding value, v. The learning entity scans the values; e_i, finds the element e_j, the closest to e, and joins it to the value v_j.

The computerized system is used as a model, enabling formalization and deduction for humans. In the computerized case, it is possible to observe a series of Expert Systems in the ascending orders of complexity, defined as follows:

Ÿ Binary case — the elements may receive only one of two values: “yes” or “no”, without the possibility of attaching two different values to the same element.

Ÿ Ternary deterministic — the elements may receive one of three values “yes”, “no”, and “maybe”.

Ÿ Ternary nondeterministic — here the same element may appear with various values that are statistically supported.

Ÿ Canonization of elements — here collection A is defined as A = {A₁, A₂, …, A_k}, where A_l (l = 1, 2, …, k represents: ; the computerized system deduces, based on the element , the corresponding value, v_l.

Ÿ Element-ranged canonization — here the collection of elements A is defined as follows: 

 (l = 1, 2, …k), A = {A₁, A₂, …, A_k,}

 is an interval of integer numbers; the computerized system deduces the corresponding value v_l, relying on elements from the interval A_j.

The following examples, graded according to their complexity and their level of uncertainty, represent the above Expert System’s classification:

Ÿ Determining correct blood transfusion;

Ÿ Teaching the Traffic Light rules in two modes:

* Deterministic

* Nondeterministic

Ÿ Identifying Deductive characters

Ÿ Recognizing Color Hues — based on the colors’ RGB ranges.

Key words: expert system; knowledge transferring; structured query language (SQL); determinism; optical character recognition (OCR)

JEL code: C630