What Are the Different Types of Retention Specialist?

Expert system tools, that is, expert system development tools, are an advanced program system or advanced programming language environment that people design for the efficient development of expert systems. Expert system tools can be divided into three types: skeleton type, auxiliary type and general type.

Building a good expert system is not easy, and usually requires the cooperation of domain experts and knowledge engineers for several months or even years. Domain experts sum up and establish the knowledge that should be placed in the expert system and establish the goals that the expert system should achieve. Therefore, they are undertaken by experts in the field to be developed with deep professional knowledge and rich practical experience. The knowledge engineer extracts the essence from the knowledge and experience summarized by the domain experts, designs the knowledge structure, selects the knowledge expression method, establishes different knowledge storage methods and reasoning mechanisms according to different knowledge expressions, and uses artificial labor such as LISP and PROLOG. The intelligent programming language implements the system prototype, and after trial and modification, it is submitted to the end user. Therefore, it is believed that if the research on practical expert systems is limited to manually developing each system from scratch using LISP or PROLOG, there will be many repetitive tasks, low efficiency, high cost, and unable to meet the needs of practical applications.

In order to transform the development of the expert system from a manual operation to a semi-automated batch production method, so as to increase the development speed of the expert system and reduce the development cost, an expert system development tool system appeared in the late 1970s, which can help domain experts directly generate their own Expert system, thereby shortening the cycle of building a practical expert system. ^[1]

The process of developing expert systems with expert system tools is shown in Figure 1.

Expert system tools can be divided into three types according to their functions: skeleton type, auxiliary type and general type.

Generally speaking, the expert system framework has the following limitations:

(1) The structure of the framework is specific to a certain area. This is because the internal structure of the original expert system is affected by its application field and has a certain field specificity.

(2) The reasoning mechanism cannot express the use of knowledge in the new domain. When the reasoning mechanism used by the expert system is far different from the solution of expert problems in a new field, the reasoning process of the expert system framework is not easily understood by the experts, making it difficult for the experts to accept the systematic reasoning ideas.

(3) Knowledge representation is not suitable for expressing knowledge structure in new fields. In a specific field, knowledge is organized around a specific environment and a specific object. For example, MYCIN is organized around context nodes; HEARSAY-II is organized around different stages of speech understanding. If the relationship between objects, concepts, and environments in the application domain is represented by the attribute inheritance structure, and the knowledge representation + attribute inheritance structure is regarded as the knowledge structure, each expert system framework can be considered to have a definite knowledge structure. Different fields may be adapted to be described with different knowledge structures, which brings difficulties to the application of the expert system framework. ^[2]

AGE system provides several expert system framework components for expert system designers to choose from. These framework components include a reverse chain framework that embodies the ideas of MYCIN and a blackboard framework that embodies the ideas of HEARSAY-II. Each framework component of AGE consists of several modules that have been carefully selected, defined, and written into modular programs. Within each framework component, the way of combining program modules has some flexibility. For example, AGE's reverse chain framework requires users to provide regular predicates that define the calculation of certainty factors. But different framework components are mutually exclusive, and expert system designers can only choose one of them at a time.

The main advantage of the AGE system is that, using the various frame components provided by the AGE, an expert system using the existing typical structure can be quickly constructed for urgent needs. In addition, because AGE provides several series of structures, its domain adaptability is correspondingly enhanced; because it provides the establishment and auxiliary function blocks corresponding to each series of structures, it makes the establishment of the expert system have some flexibility and domain specificity. . ^[2]