The Representation of Knowledge2.1

The Representation of Knowledge
2.1 INTRODUCTION
In this chapter we will discuss some of the commonly used representations of knowledge for expert systems. Knowledge representation is of major importance in expert systems for two reasons. First, expert system shells are designed for a certain type of knowledge representation such as rules or logic. Second, the way in which an expert system represents knoeledge affect the development, efficiency, speed, and maintenance of the system. In Chapter 3, we will discuss how inferences are made.
2.2 THE MEANING OF KNOWLEDGE
Knowledge, like love, is one of those words that everyone knows the meaning of, yet finds hard to define, like love, knowledge has many meanings. Other words such as data, facts, and information are often used interchangebly with knowledge.
The study of knowledge is episstomology (Angeles 81). It is concerned with the nature, structure, and origins of knowledge. Figure 2.1 illustrates some of the categories of epistemology. Besides the philosophical kinds of knowledge expressed by Aristotle, Plato, Descartes, Hume, Kant, and others, there are two special types, called a priori and a posteriori. The term a priori comes from the Latin and means “that which precedes”. A priori knowledge comes before and is independent of knowledge from the senses. As an example, the statements “everithing has a cause” and “all triangles in a plane have 180 degrees” are examples of a priori knowledge. A priori knowledge is considered to be universally true and cannot be denied without contradiction. Logic statements, mathematical laws, and the knowledge possessed by teenagers are examples of a priori knowledge.
The opposite of a priori knowledge is knowledge derived from the senses, or a posteriori knowledge. The truth or falsity of a posteriori knowledge can be verified using sense experience, as in the statement “the light is green”. However, because sensory experience may not always be reliable, a posteriori knowledge can be denied on the basis of new knowledge without the necessity of contradictions. For example, if you saw someone with brown eyes, you would believe that person’s eyes were brown. However, if you later saw that person removing brown contact lenses to reveal blue eyes, your knowledge would have to be revised.
Knowledge can be further classified into procedural knowledge, declarative knowledge, and tacit knowlwdge. The procedural and declarative knowledge types correspond to the procedural and declarative paradigms dscussed in Chapter 1.
Procedural knowledge is often referred to as knowing how to do something. An example of procedural knowledge is knowing how to boil a pot of water. Declarative knowledge refers to knowing that something is true or false. It is concerned with knowledge expressed in the form of declarative statements such as “Don’t put your fingers in a pot of boiling water”.


EPISTOMOLOGY



PHILOSOPHIC A PRIORI A POSTERIORI
THEORIES KNOWLEDGE KNOWLEDGE



ARISTOTLE
PLATO
KANT
LOCKE
MILL
Figure 2.1 Some Categories of Epistemology
Tacit knowledge is sometimes called unconscious knowledge because it cannot be expressed by languge. An example is knowing how to move your hand. On a gross scale, you might say that you move your hand by tightening or relaxing tighten or relax the muscles and tendons? Other examples are walking or riding a bicycle. In computer systems ANS is related to tacit knowledge bacause normally the neural net cannot directly explain its knowledge, but may be able to if given an appropiriate program (see Section 1.14).
Knowledge is of primary importance in expert systems. In fact, an analogy to Wirth’s classic expression
Algorithms + Data Structures = Programs
For expert systems is
Knowledge + Inference = Expert Systems
As used in this book, knowledge is part of a hierarchy, illustrated in Figure 2.2. at the bottom is noise, consisting of items that are of little interest and that obscure data. The next higher level is data, which are items of potential interest. Information, or processed data that are of interest are on the third level. Next is knowledge, which represents very specialized information. In Chapter 1, knowledge in rule-based expert systems was defined as the rules that were activated by facts to produce new facts or conclusions. This process of inferencing is the second essential part of an expert systems. Reasong is generally used in human thinking.

META-
KNOWLEDGE
KNOWLEDGE
INFORMATION
DATA
NOISE

Figure 2.2 The Hierarchy of Knowledge
The term facts can mean either data or information. Depending on how they are written, expert systems may draw inferences using data or information. Expert systems may also (1) separate data from noise, (2) transform data info information, or (3) transform information into knowledge.
As an example of these concepts, consider the following sequence of 24 numbers :
137178766832525156430015
Without knowledge, this entire sequence may appear to be noise. However, if it is known that this sequence is meaningful, then the sequence is data. Determining what is data and what is noise is like the old saying about gardening, “a weed is anything that grows that isn’t what you want”.
Certain knowledge may exist to transform data into information. For example, the following algorithm processes the data to yield information.
Group the numbers by twos.
Ignore any two-digit numbers less than 32.
Substitute the ASCII characters for the two-digit numbers.
Application of this algorithm to the previous 24 numbers yields the information
GOLD 438+
Now knowledge can be applied to this information. For example, there may be a rule
IF gold is less than 500
And the price is rising (+)
THEN
Buy gold
Although not explicitly shown in Figure 2.2, expertise is a specialized type of knowledge that experts have. Expertise is not commonly found in public sources of information such as books and papers. Instead, expertise is the implicit knowledge of the expert that must be extracted and made explicit so it can be encoded in an expert system. Above knowledge is metaknowledge. One meaning of the prefix meta is “above”. Metaknowledge is knowledge about knowledge and expertise. An expert system may be designed with knowledge about several different domains. Metaknowledge would specify which knowledge bases about car repair of 1988 Chevrolets, 1985 Fords, and 1989 Cadillacs. Depending on what car needed repair, the appropriate knowledge base would be used. It would be inefficient in terms of memory and speed for all of the knowledge bases to be working at once. In addition, there could be conflicts as the expert system tried to decide the applicable rules from all knowledge bases at once. Metaknowledge may also be used within one domain to decide which group of rules in the domain is most applicable.
In a philosophical sense, wisdom is the peak of all knowledge. Wisdom is the metaknowledge of determining the best goals of life and how to obtain them. A rule of wisdom might be
IF I have enough money to keep my spouse happy
THEN I will retire and enjoy life
However, due to the extreme scarcity of wisdom in the world, we shall restrict ourselves to knowledge-based systems and leave wisdom-based systems to politicians.