In fact, there are many artificial intelligence programming languages corresponding to various knowledge representations. According to the corresponding knowledge, it is different. Generally can be divided into the following categories:
1. corresponds to the language of knowledge representation of production rules. For example, 1 977 OPS (official production system) developed by C L Forgy of Carnegie Menon University in the United States was used to develop DEC's expert system X 1/XCON.
2. The language corresponding to the knowledge representation of logical formulas. A logic language that has been widely used is PROLOG. It was invented by Alain Colmerauer of the University of Marseille in France in 1 970.
3. The language corresponding to the knowledge representation of the framework or semantic web. This is a so-called object-oriented language. One of the representative languages is Smalltalk. It was first created in 1980, and then several versions were developed, usually marked by the release year. For example, the version of 1 is called Smalltalk-80 or something.
4. Languages corresponding to functional knowledge representation or functional programming style. First, it was put forward by J. Backus, a computer scientist, in a famous paper published in 1978 that won the Turing Prize. The title of this paper is: "Can programming get rid of Neumann style? The functional style of programs and their algebras ". Although its functional programming language is perfect in theory and based on a solid mathematical foundation, it is difficult to realize on a conventional computer. On the contrary, LISP (List Processing), which was first developed by john mccarthy of Massachusetts Institute of Technology in the late 1950s and early 1960s, is still widely used to write artificial intelligence applications, especially to develop expert systems. The feature of functional language in explaining the execution mechanism is recursive reduction from the innermost layer to the outer layer, and each reduction is a process of "acting" a function on its independent variables to obtain the function value. Therefore, it is also called "application" language.
One common feature of artificial intelligence programming languages is that these languages are designed independently for the problems to be solved, combined with knowledge representation, and completely divorced from the Neumann structural characteristics of contemporary computers; They are at a higher level of abstraction than process-oriented high-level programming languages. Therefore, programs written in these languages, whether interpreted or compiled, are usually inefficient in modern computer environment. Especially when the program is huge and complex, it will waste a lot of system resources (mainly referring to the time and storage space occupied by the processor), which will make the system performance drop to an intolerable level.