Al-Huda
Foundation, NJ U. S. A
the Message Continues ... 8/121
Newsletter for September 2011
Article 1 - Article 2 - Article 3 - Article 4 - Article 5 - Article 6 - Article 7 - Article 8 - Article 9 - Article 10 - Article 11 - Article 12
Islamic
Perspectives on Knowledge Engineering
S. Imtiaz Ahmad
Abstract This paper provides Islamic perspectives on the recent developments in knowledge engineering. The paper starts by introducing the concepts relevant to knowledge engineering. The use of the science of knowledge, and the scientific method for knowledge based systems is described and discussed. The role of knowledge engineering is defined. Relationship of knowledge engineering to the study of human mental faculties is described in intelligence and psychology. Finally, a brief description of the historical developments and current trends is presented. I. Introduction The purpose of this research is to find answers to some basic questions regarding the human mental processes that occur in dealing with knowledge, and the manner in which this knowledge is organized for innovation and problem solving. In reading the Qur'an and literature on Hadlth of the Prophet, the basic sources of information on Islam, one finds a great deal of emphasis on knowledge and the special position given to those with knowledge. One, therefore, wonders as to why the Muslims, professing Islam, do not appear to be the leaders in knowledge. In this case however it is not only necessary to understand why that may be the case but also to search for ways to correct this situation. Research in human problem solving has resulted in paradigms about the working of the human mind. Advances in computers and the use of computers in problem solving, in particular the use of the techniques based on artificial intelligence, are having further impact on these paradigms. It is to be noted that while these paradigms may give a rational explanation to what we perceive through our senses, they do not necessarily represent the ultimate reality of the human mind. Knowledge comes from learning about things, and it requires mental apprehension or cognition. The process of learning consists of using existing knowledge, gaining new knowledge, organizing, and storing the new and old knowledge. Stored knowledge is recalled and used in responding to the events in the environment. Our present understanding of how the human mind organizes knowledge points to the following two characteristics: what the mind stores is a refined form of received information, and it also retains the context of this information. The main topic of the Qur'an is also the human mind. In particular, it deals with the question of how the humans do use or should use their mind in responding to the events in the environment. Based on our current understanding, it appears that the style of Qur'anic descriptions is well suited to human cognitive skills, e.g., • Details are relevant to a context. • Reasoning is goal oriented. • The same subject is presented using several alternate perspectives. • The message is conveyed through patterns of things or parables, called amthal-ul Quran. • Positive as well as contrary-to-positive templates of behavior, Mujibat al Falah and Mujibat al Khusran, are presented. Phillip Selznick, in his book, Leadership and Administration, argues (15) that the human values are not usually transmitted through formal written procedures. They are more often diffused by softer means: specifically the stories, myths, legends, and metaphors that we have already seen. This argument is based on the impact on human mind produced by a certain style of stating facts, and it conforms well to the cognitive skills stated above. Some examples of the concepts and terms that the Qur'an uses about knowledge are: • Being aware of, or having cognizance of (dirayah). • Insight, literary and spiritual. • Reasoning and rationalizing with facts and rules (Qur'an 29:49). • Contrary to guesswork or conjecture (Qur'an 7:7). • Learning and discovering truth (Qur'an 22:54). Knowledge engineering deals with building systems based on the knowledge of someone who is well experienced in dealing with the events of some domain of application. Events generate stimuli and require responses. One must properly recognize the event, and then generate an appropriate response; this process is called problem solving. In solving a problem, one must have access to what may be previously known, retrieve relevant data and rules, and apply this knowledge effectively. This may be described as a goal seeking process. Given some premise and some desired goal, one may use the knowledge to move forward from the premise toward the desired goal until that goal is reached. Alternatively, one may use the knowledge to move backward from the goal to the premise in order to establish that the goal can be reached from the premise. In either case, one goes through several intermediate steps, and collectively these steps constitute a chain of thought. Reasoning from the premise to the goal is called forward chaining, and reasoning from goal to the premise is called backward chaining; the choice generally depends on the situation. Many people do not possess the knowledge that may be required to solve problems in a given area, or they are unable to use effectively the knowledge that they have. Those who do, are known as experts. It takes a human being many years, possibly decades, to become an expert in some area. Direct use of an expert's knowledge is limited by the possibilities of personal contacts. However, if one can acquire successfully the knowledge of how an expert solves problems then it can be put to widespread use. Furthermore, if one can successfully transfer the expert's knowledge to a machine, then the access and use of this knowledge can be increased manifold. Moreover, one is now able to exploit the inherent capabilities of the machine to store vast amounts of information, recall it when needed, and put it to use at lightning speed. II. Knowledge Engine A mechanism for storing and organizing facts and rules from known situations, and using them for resolution of new situations is called a knowledge engine. Designed properly, a knowledge engine can unleash the problem solving power contained in knowledge. Traditionally, the human mind has served as the knowledge engine. It is fueled by the stimuli from the environment, uses existing knowledge to process information, solves problems, acquires new knowledge in the process, organizes and updates existing knowledge, and generates information leading to the creation of new knowledge. Before the industrial revolution, tools for enhancing the mechanical abilities of humans were rather limited. With the industrial revolution came the steam, oil, electric, hydro, and nuclear powered engines which allowed the human race to alter the physical environment for its purpose. The changes affected the quality of life dramatically. With powered machines, it became unnecessary for human beings to exert their body to lift heavy leads, walk long distances, or endure harsh climates. Moreover, the power of one such engine could out perform a large number of human beings. The power machines of the industrial revolution gave vast amounts of physical power to the human beings for their use. However, in order to keep up with these machines, the human beings were often required to perform repetitive tasks. Also, as the power and operational capabilities of the machines were increased, they replaced an increasing number of human beings in the work environment. In a similar fashion, knowledge engines, also called information machines in a more general sense, are bringing another kind of revolution. One may call it an information revolution. A mechanical machine can easily exceed the physical capacity of hundreds of human beings, particularly for tasks that do not require physical dexterity. An information machine, likewise, can exceed the mental capacity of hundreds of human beings for tasks that do not require mental adroitness. Furthermore, a human being can make the information machine to act as an intelligent assistant in his work, allowing him to be more productive mentally, much the same way that the mechanical machine can make him more productive physically (1, 16). III. Of Mind and Machines A human being has physical faculties of force and motion, sensory faculties of seeing, hearing, touching and smelling, as well as the mental faculties. Examples of mental faculties are perceiving on seeing, discerning on hearing, and reasoning with facts and rules. Other sensory faculties such as the touch and smell also produce messages to be appropriately processed by the mental faculties. The information processing machine has the potential to enhance one's mental faculties. Those who use these machines can enhance their mental faculties, produce more goods and exert greater power and control in society. Unless the overall opportunities grow at a faster rate, the potential for others will continue to diminish. Examples of these are given below in items a to d. An information machine consists of a computer, a knowledge base of data and models, and a mechanism for controlling the operations. Control of operations includes the selection and application of data and models relevant to a situation. There is indeed no doubt that the information machine, when used properly, serves to increase human productivity. The machine helps to organize and store the information needed to generate responses to events in the environment, select required information from thousands of stored pages, and scan for specific items from hundreds of pages in a matter of seconds. Machines have a potential for benefit as well as detriment. Benefits come from one's ability to enhance the mental processes. Detriment lies in letting the machine take over one's normal mental processes. With proper use, an information machine may be visualized as a mind expander. Because of the unlimited potential that these machines offer, uses and abuses of information technology are likely to be far more profound than those brought about by the industrial revolution. Consider the following examples: a . Planning, control, and review of complex business and government enterprises requires vast amounts of rapidly accessible information. Those who have information technology can run the enterprise productively, increase their ability to produce more, enhance quality, and use less resources. They can, thus, dominate their competitors quantitatively, qualitatively, and intellectually. b. An architect, an engineer, or an accountant may be able to do high quality work better than ten architects, engineers, or accountants, respectively by using the information technology. The other nine architects, engineers, or accountants thus replaced must adapt or be eliminated. c. The rapid changes occurring in information technology create rapid obsolescence, and new learning requirements. There may be many who are not educated enough to adapt to this rapid change. d . Products of information technology which successfully model human mental faculties may gradually take over the work in many areas of human services, with possible intimidating situations. Each example points to the benefits for those who can make timely and effective use of information technology. However, the same technology becomes detrimental to those who are unable or unwilling to deal with it effectively. IV Knowledge Engineering Concepts The basic material of knowledge engineering is information, in raw and refined form. This information consists of descriptions of object types covering their explicit and implicit attributes and instances. Alternatively, we may say that an object type is described by some `relevant' attribute names, whereas a specific instance of an object defines attribute values. Frequently, one may use the term object to refer to object name, attributes to refer to attribute names, and values to refer to instances. As an example, consider a patient in a hospital. Here, one of the object types is patient, and the other is hospital. For the patient, the relevant attributes may be name age, symptoms, history, etc. A specific instance may be John Adams, 33, fever, none, etc. The choice of attributes for an object depends on what information needs to be represented. Furthermore, one must consider how this information should be structured to properly satisfy the requirements of the applications dealing with the objects. The extent to which an object is described depends both on our understanding about the object and the assumed context of the application. The process of abstracting the attributes of an object may be difficult, particularly in the absence of any prior experience with it. Furthermore, the notions of how an object type should be described may change with time due to changes in our understanding about the object, or changes in the application context. One must also consider ways of collecting attribute values, and keeping them current, possibly maintaining a history of these values. For example, the current value of weight attribute may be relevant for a patient but so may be the previous history of the weight values, Le, how the weight has been changing between checkups. It is not always possible to define precisely what attributes may describe an object adequately unless one is an expert on it. Previous knowledge about the object can be quite helpful in making the right choice. Consider the first lesson in knowledge engineering given to Adam by God, and described in the Qur'an as: "And taught Adam the names of things." [Qur'an 2:31-33.]. It appears that in this lesson a process of synthesis of knowledge was in the making: patterns were in motion, and recognition was in action. This was the first phenomena that involved the human mind in abstracting the attributes and assigning names to things based on those perceived attributes. The Qur'an describes knowledge and the principles and tools of knowledge engineering as: • Ilm al-Yaqin [Qur'an 102:5], certainty or knowledge gained from reasoning and inference. • Ayn al-Yaqin [Qur'an 102:7] , certainty or knowledge gained from sight (from the senses), and • Haqq al-Yaqin [Qur'an 69:51], certainty or knowledge that is absolute in truth, not subject to alternation from knowledge received through sense perceptions, reasoning, or inference. The first two items are related to the knowledge that is acquired, and the third item points to the knowledge revealed to mankind through the ages. Description of an object is not simply as to what it is, but also what capabilities it may have. The capabilities describe the operations the object permits, as well as those it can perform, resource requirements, and constraints. An object may be manipulated by some objects, and it may manipulate some of them. The extent of an object description, and the ability to acquire instances of this description, determine the scope of the responses which may be generated when events related to the object occur in the application environment. All these considerations are relevant to engineering useful knowledge about an object. The product of knowledge engineering is a system consisting of a knowledge base structure, an interface for knowledge acquisition and user queries, and a mechanism for activating the knowledge base in order to generate responses all residing in a special or general purpose computer. Knowledge engineering deals with the concepts, tools, and techniques for describing the objects, structuring the description for acquiring and maintaining information, and developing mechanisms for sequencing of the operations [6, 7, 18, 21]. It also deals with the mechanisms for creating, mutating, and deleting the objects. The processor in the computer provides the raw power, fueled by the data and logic components of the knowledge base, to work as a knowledge engine. Speaking broadly, and sounding somewhat futuristic, one may define the goals of knowledge engineering as: • Creating intellect from knowledge, i.e., creating a machine that could reason as a philosopher, offering new insights into historical and contemporary events. • Creating mind inside matter, i.e., creating a machine capable of independent thought. We will elaborate these goals further in the sections that follow. V Role of Knowledge Engineer A knowledge engineer is responsible for creating a mirror image of a particular reality, i.e., creating an authentic model of what exists in the application domain. This work requires discovery of what the reality is or how it is perceived, developing a representation consistent with the events and responses in the real world, and maintaining the integrity of the representation. In order to perform this task, the knowledge engineer must understand the science of knowledge, and the manner of its application. Knowledge comes through observations, reasoning, and reflection. There are two categories of knowledge- axiomatic and empirical. Axiomatic knowledge deals with the possibility of possible things, and the impossibility of impossible things. Given an event, and axiomatic knowledge about it, one may describe a definite response. Empirical knowledge, on the other hand deals with observation and experimentation. Given an event, and only empirical knowledge about it, one may develop a response based on experience. In performing an analysis of the situation which is to be modeled, the knowledge engineer is required to use all sources of information, to discern specifics of the application domain, and to describe the knowledge thus gained. Generally, the knowledge engineer, or in this phase of the work one may call him the knowledge analyst, is not the creator or user of the knowledge in the application domain. He must refer to those who can validate his knowledge of the application domain. This requires tools and techniques of communication, and their use in a manner which encourages the vocalization of pertinent information. The purpose of the validation process is to remove ignorance about the application domain, and generate the knowledge for modeling and representation of the reality. Systems built on ignorance about the domain of application either fail completely, or perform very poorly. However, at times it may be necessary to build a system based on incomplete knowledge. This deficiency may be overcome by a mechanism which explains what knowledge was used, and how it was used in generating the response to some event. In this case, the user must have the knowledge to assess the validity of the response in a given situation. The explanation facility also indicates the need for further knowledge acquisition whenever it becomes necessary. Most application domains are dynamic in nature, i.e., data values are affected by aging, and the applicable policies are affected by changes. The knowledge engineer plays a key role in maintaining system integrity with time. The internal design of a knowledge engine, or knowledge based system, determines its space and time characteristics. One may assume that the purpose of the system is to augment human capabilities, and increase productivity of the operations. The knowledge engineer is, therefore, responsible for providing the facilities for the users to interact with the system. This interaction should allow the users to maintain their normal intellectual thought processes. The subservience, if there is to be one, should be of the system to the user, and not the other way around. VI. Knowledge Based Systems and Artificial Intelligence One may describe knowledge as a collection of facts and heuristics. Facts represent that part of knowledge which is widely shared, publicly available, and generally agreed upon by experts in a field. Heuristics, on the other hand, represent that part of knowledge which is mostly private, little discussed rules of plausible reasoning, good judgment, and good guessing. Knowledge based systems store facts and heuristics for making inferences about situations. If the facts and heuristics normally used by an expert are acquired and properly represented in a system, then such a system is called a knowledge based expert system [5, 7], or simply an expert system. Artificial intelligence is the study of mental faculties through the use of computational models [2]. If what the brain does can be modeled as a computation then the work in artificial intelligence will successfully duplicate the human mental faculties. For example, the models of human mental faculties in vision and natural language are useful in building systems for machine vision [9, 17] and machine processing of natural language [4]. Also see examples of such applications in [3, 10, 12, 13, 14]. All humans, not just the experts, have these faculties. The tools and techniques of artificial intelligence are used in building intelligent systems based on human mental faculties. The work in psychology, dealing with the study of human mind, has influenced the direction of work in artificial intelligence. Looking at what the psychologists have to say, about the human mind, may help one better understand the current work and future trends in artificial intelligence. According to the theory of behaviorism in psychology, all human behavior can be described in terms of a cause and effect relationship between the stimuli from the external events and the responses. Once this relationship is understood and described in the form of a stimulus-response mechanism, it then becomes possible to predict and control human behavior. First definitive work on this subject was published by Watson who said [19]: Psychology as the behaviorist sees it is a purely objective, experimental branch of natural science. Its theoretical goal is the prediction and control of behavior. Introspection forms no essential part of its methods, nor is the scientific value of its data dependent upon the readiness with which they lend themselves to interpretation in terms of consciousness. The behaviorist, in his effort to get a unitary scheme of animal responses, recognizes no dividing line between man and brute. Behaviorism however, has not succeeded in producing a theory of behavior that is applicable in all situations [8]. Nonetheless, it continues to play a major role in situations requiring behavior modification. Cognitive psychology introduces the notion of thinking, i.e., people interpret the external stimuli by a thought process in order to produce a response. The passive cause and effect relationship advanced by the behaviorist is, therefore, not applicable to human behavior in all situations. The cognitive psychologist distinguishes the human from the other animals. It is, however, not clear whether the human ability to interpret, as seen by the cognitive psychologist, allows for the possibility of directing ones actions, known as free-will, without constraint by necessity or fate. One may summarize the cognitive psychology model of the human mind in terms of the following features: • The mind has operationally definable mediators (Logical Behaviorism), • The mind has a central mechanism for mediators (Central Cognitive Process -Subprocesses), • Central mechanism is not reducible to behavioral or peripheral terms (Contemporary Cognitive Behaviorism, or Information Processing Approach). Contrast this model with the classical references to the human mind as the tabula rasa, i.e., the human mind is a blank tablet at birth, and the sense experience is the only source of knowledge. In the Qur'an the words from God are: "When I fashioned him (in due proportion) and breathed into him My Spirit." [Qur'an 15:20]. Breathing of God's spirit implies giving the faculty of God-like knowledge and will. Rightly used, it distinguishes man from other creatures. The Qur'an, therefore, invalidates the common interpretation of tabula rasa. The cognitive psychologist seem to have come to the same conclusion in their own studies of the human mind. We may further add that the sense experience is the external source of knowledge, and this knowledge may be internally manipulated in ways that is not always predictable or reducible to behavioral or peripheral terms. B.F. Skinner in his book, Beyond Freedom and Dignity, says: "We are all simply a product of the stimuli we get from the external world. Specify the environment completely enough and you can exactly predict the individual's actions:" [15]. This can only be true however, if the individual does not properly use the free-will given to him by God and allows himself to be blindly shaped by the changes in the environment. Bruno Bettelheim in his book, On the Uses of Enchantment, sounded a positive challenge for the human mind when he said: "If we hope to live not just from moment to moment, but in a true consciousness of our existence, then our greatest need and most difficult achievement is to find meaning in our life." The fields of psychology and artificial intelligence will continue to crossfertilize. Of course, a theory in psychology about the human mind does not mean that the mind actually works that way. It is important to make this distinction. Furthermore, a machine built on models of the human mind is just that, i.e., it exhibits intelligent behavior but does not necessarily duplicate human intelligence. With increasing intelligence in the model, the reality may still be distinctively different. The work in artificial intelligence does not claim that its goal is to produce methods which duplicate exactly those of the people [2]. Its goal is to build systems which exhibit intelligent behavior, solving problems in ways that resemble those of the humans. Again, it is important to make a distinction between a machine which exhibits intelligent behavior and an intelligent human being. It is necessary to keep the humans ahead of the machines, retaining the challenge to improve the machines. VII. Development of the Human Mind Our interactions with the environment are permanently recorded in our mind via the sense perceptions. Roger Penrose, a well known mathematician once said: "The world is an illusion created by conspiracy of the senses [15]. Our mind stores the information received from the senses in a variety of ways. All of this information can be recalled under appropriate conditions. This is a working premise of knowledge engineering. The stored information appears to not depend on the language in which the information is transacted. Consider now what is said in the Quran about the sense perceptions: • That day shall We set a seal on their mouth but their hands will speak to Us, and their feet will bear witness, to all they did. (Qur'an 36:65). • Their hearing, their sight, and their skins will bear witness. (Qur'an 41:20). • On the day when their tongues, their hands, and their feet will bear witness against them as to their actions (Qur'an 24:24). Thus, the skin sends signals (speaks) to the brain from the senses of touch, taste and smell, as does the eyes on seeing and the ear on hearing. The mind can recall these signals and vocalize them in any spoken language. The term nafs (soul) is used in th Quran in a manner cognate to the human mind. Consider the following quotations: • Who created you from a single person. Quran 4:1). • No soul can believe except by the Will of God. (Qur'an 10:100). • And the soul and the proportion and order given to it. Quran 91:7). • Do they reflect not in their own mind. (Quran 30:8) • Soul prone to evil. Quran 12:53). • Self reproaching soul. (Qur'an 75:2). • Righteous (at rest and satisfied) soul. Quran 89:21). The above characterizations in the Qur'an point to various aspects of the development of the human mind. VIII. Trends in Knowledge Engineering Earlier work in the use of computers for knowledge engineering was limited to areas of axiomatic knowledge, facts consisting of data and computations. In scientific and business applications, many situations allowed descriptions of planned responses to events in the environment. These early systems were called information processing systems, or simply information systems. Gradually, the developments in information technology and the understanding of its potential in human productivity, resulted in emphasis on building systems to support decision making [ll]. Often, the decision making situations cannot be described fully in terms of cause and effect relationships. The system, therefore, consists of data, models, and interfaces to interact and produce ad hoc responses which the people could analyze and choose for making decisions. These systems are called decision support systems. In many situations decisions are based on plausible reasoning, more a matter of good judgement on the part of an expert. These expert's knowledge may be represented using the tools of artificial intelligence. Systems based on this knowledge, detailed and specific to a domain of application, are called expert systems. All of the above mentioned systems may be considered as instances of knowledge based systems, created to serve the potential users. Acknowledgements: The author gratefully acknowledges the contribution of Nejma Natalie Heisler who gave many suggestions regarding the contents, and the style of presentation. REFERENCES 1. Ahmad, S.I., Basic, M., Dutta, S.P, Lashkari, Ragab, M., Raouf, A., Renaud, J. and Shklov, N., "Production Research as a Means of Improving Productivity; Proc.l VIIth. International Conference on Production Research, University of Windsor, Windsor, 2 vols., 1983. 2. Charniak, E. and McDermott, D., Artificial Intelligence, AddisonWesley, 1985. 3. Duda, R.O., Gaschnig, J.G. and Hart, P .E., "Model Design in the PROSPECTOR Consultant System for Mineral Exploration; D. Michie, (ed.), Expert Systems in the Microelectronics Age, Edingburgh University Press, pp. 153-167, 1980. 4. Erman, L.D., "Hearsay II Speech Understanding System: Integrating Knowledge to Resolve Uncertainty; Computing Surveys, 12(2), pp. 213-253, Feb. 1980. 5. Harmon, P and King D., Expert Systems, John Wiley, New York, 1985. 6. Hayes-Roth, R, "Knowledge-Based Expert Systems; Computer, 17(10), pp. 263-273, Oct. 1984. 7. Hayes-Roth, F , Waterman, D. and Lenat, D., Building Expert Systems, Addison-Wesley, 1983. 8. Hothersal, D., History of Psychology, Random House, NY, 1984. 9. R. Kuppala, "Automatic Laboratory Scale Production Cell with Vision; Proceeding of the 14th. International Symposium on Industrial Robots and the 7th. International Conference on Industrial Robot Technology, Gothenburg, Sweden, pp. 183-188, October 1984. 10. Lindsay, R.K., Buchanan, B.G., Geigenbaum, E.A., and Lederberg, J. Applications of Artificial Intelligence for organic Chemistry: 7yte Dendral Project, McGraw-Hill, 1980. 11. Leigh, W .E., Decision Support and Expert Systems, South Western Publishing Co., Cincinnati, 1986. 12. Martin, W .A. and Fateman, J., "The Macsyma System; Second Symposium, Symbolic and Algebraic Manipulation, pp. 59 75, 1971. 13. McDermott, J., "RI: The Formative Years; AI Magazine, 2(2), pp. 21-29, 1981. 14. Miller, R. People, H., and Myers, J., "Internist-1, An Experimental Computer-Based Diagnostic Consultant for General Internal Medicine" New England Journal of Medicine, 307, pp. 468-476, 1982. 15. Peters, TJ. and Waterman, Jr., R.H., In Search of Excellence, Warner Books (Harper & Rows), NY, 1982. 16. Raiouf, A. and Ahmad, S.I., (eds.), Flexible Manufacturing: Recent Developments in FMS, Rotoucs, CAD/CAM, CIM Elsevier, Amsterdam, 1985. 17. Rogers, H.P "Helping Machines See; Canadian Datasystems, 18(2), pp. 38-39, February 1986. 18. Waterman, D.A., A Guide to Expert Systems, Addison-Wesley, Reading, 1984. 19. Watson, J.B., "Psychology as The Behaviorist Sees It; Psychological Review, 20, pp. 158-177, 1913. 20. Winston, P .H., Artificial Intelligence, Addison-Wesley, Reading, 1984. 21. Wos, L. Overbeek, R., Lust, E., and Boyle, J., Automated Reasoning, Prentice Hall, Englewood Cliffs, 1984. Excepted from: Islamization of Attitudes and Practics in Science and Technology, by S. Imtiaz Ahmad |
HOME - NEWSLETTERS - BOOKS - ARTICLES - CONTACT - FEEDBACK
DISCLAIMER:
All material published by Al-Huda.com / And the Message Continues is the sole responsibility of its author's).
The opinions and/or assertions contained therein do not necessarily reflect the editorial views of this site,
nor of Al-Huda and its officers.