A dumb information system in the information supply chain can provide data that is often difficult for the customer to interpret and use. To help in this regard, machine intelligence based on some learning rules is introduced in this chapter. Architecture of the knowledge base and the rule base are explained. The acquired knowledge from the different sources is to be consolidated.
TopIntroduction
The last chapter deals with the identification of patterns in the data. The occurrence of the patterns in the data takes it to a higher level of organization, clearly indicating it is not junk. Based on a chain of reasoning and inferences, these data patterns provide a solution for complex problems. Here, the architecture to build a system that provides expert solutions for the domain specific problems is discussed.
In a traditional computing system, there is less scope for providing the solutions for the problems based on the previous learning. The solutions are hardcoded with a rigid program language and remain the same always. On the other hand, in an expert system, the solution is coded in the database rather than being provided by a program. This architecture is advantageous in several ways.
In an expert system, the information is modeled at a higher level of the abstraction. To build the model resembling the human reasoning, support from different tools and symbolic languages is sought. The language used for modeling supports Rule-based programming, where a set of rules are used to represent heuristics or actions under s certain context or situation. A Rule has two portions:
- 1.
An if portion where a series of conditions or patterns or stimuli are specified. These patterns are to be matched with the input pattern to excite the appropriate condition.
- 2.
A then portion where the effects or the consequences are described when the if portion turns out to be true.
The usage of a neural network for pattern matching has been given in the chapter on intelligent information processing. The architecture of expert systems with the organization of data in to abstraction levels in place is the topic of discussion here.
Expert systems are useful for getting specific solutions for the problems in domains as diverse as mathematics, medicine, business, law, military, education etc. Although the knowledge stored in an expert system is specific to a domain, it can provide solution for any problem with in the domain. In this chapter, the components of the expert system are discussed in detail. The architecture of the expert system in turn determines the design of the applications calling the same. The different services that need the help of an expert system are e- governance, portals, knowledge management (Liebowitz, J. & Chen, Y, 2001), etc. Generally, the queries focus on:
- •
Diagnosis: Used to find the faults in the data when it is difficult to carryout manually. For example, design of a million gate integrated circuit.
- •
Data Interpretation: When it is difficult to analyze a large volume of data, for example, geological data, the help of an expert system is sought to find the patterns and interpret the data.
In an expert system, the solution provided depends more on the problem type than on the domain of the problem. In its simple form, it takes the queries from the user or the invoking applications, provides the required information as well as clarification required to solve the problem or help in the decision making (Power, D. J, 2002). Often it is iterative providing more precise answer in each of the iterations. Thus it is much powerful than an ordinary search engine.
An expert system is designed to address the requirements of a specific domain with a much focused objective. This is because they perform well and maintain the knowledge base of a single domain when they have a narrow focus rather than maintaining information on all the areas that are generally not requested by the users of the system.