Expert Systems with Applications - Literature review Example

Expert Systems with Applications By Date Table of Contents Table of Contents 1 Abstract 3 Introduction 3 Literature Survey 5 Generic Design of an Expert system 7 Applications of Expert systems 10 Conclusion 13 Work Cited 15 List of Figures Figure 1: Architecture of Expert System 8 Abstract Now days, domain expertise is needed in every domain to get the proper understanding of the domain and to use domain information for the purpose of solving domain related problems. But there are some problems in accessing the domain expert every time when domain information is needed. And also there is secrecy of real experts and if they are available, it is difficult for a common man to interact with them. With the increasing computing power and changing technologies, it is now possible for computer systems to replace the role of human experts; expert systems have been developed for this purpose. Expert systems use the capabilities of computers and domain experts and techniques of artificial intelligence, in form of knowledge base and inference engine and provide its user with a quick and precise decision or solution about a problem in a particular domain. Development of an expert system is pretty mature idea now and expert systems are providing their services in many domains such as agriculture, medical, law enforcement etc. This paper explores the methodologies to design an expert system, state of the art and some important applications of expert systems. Keywords: Artificial Intelligence, Expert System, Knowledge Base, Inference Engine Introduction Rapid advancements in technologies have enabled computers to solve scientific and social problems based on the knowledge about the problem and making inferences, that is the application of an expert system. Using the techniques of AI, an expert system provides domain specific advice to its users - by incorporating sufficient knowledge about the problem domain to reach a level of performance equivalent to that of a human expert. These expert systems represent the expert’s knowledge about the domain in form of data or rules within the computer, in form of a database, which can then be used when needed to solve problems (Hou and Fan) and (Laudon and Laudon). With the aim of supporting and helping non-experts in solving some problems, an expert system comes up with a number of components that may help in producing expert advice about a particular domain specific problem. These generic components may include the knowledge base, Inference Engine and the user interface. The knowledge base contains the knowledge that is usually acquired from a domain expert. Knowledge is normally presented in the form of knowledge rules. The inference engine is basically the decision maker that interprets and manipulates the rules found in the knowledge base to make conclusion based on user query. The user interface is basically the presentation of the system to the user that allows the user to query the system and receive the expert’s advice in result of those queries. In addition, many expert systems also provide explanation about the purpose of the question and justifications for the produced advice (Wai, B and Rahman) and (Hutchinson and Sawyer). Expert systems are providing their services in various application areas such as agriculture, education, environment, law enforcements, medicine etc. The main objective is to produce relevant data and information for consultations and solving domain specific problems with the help of that information. Structure of the paper is as follows: next section explores the literature about ES’s, and then a generic architecture for ES is discussed. After that some applications of ES’s are highlighted and last section summarizes the conclusions. Literature Survey The development of an expert system is not a novel idea and a lot of research work has already been done in this domain. A large number of researchers have contributed their part by application of an expert system in a particular domain (Hou and Fan) have contributed their part in expert system’s research by presenting the idea of a virtual expert system. The idea is basically presented to resolve the difficulties in knowledge acquisition; and the problem that an expert system cannot work alone but still requires the help of a domain expert so that its result may be used effectively. Authors have suggested the use of integration of knowledge ability (IKA) theory to resolve the scientific or social issues that may hinder knowledge acquisition, developing a network of virtual team of experts that may allow the achievement of IKA for experts’ team using the theories of IKA in a certain way. The idea is expected to establish a high standard expert’s innovation platform, building valuable modes and mechanisms of IKA, selection of the members of virtual expert’s team flawlessly for complex scientific research projects. According to (Mizoguchi and Motoda), certain research work has been done regarding the application of an expert system in electric power and gas industries. From 1984 to 199 1, MIT1, an organization, sponsored a national Man-Machine Systems project for research in nuclear power plant operator assistance. This project involves important work on expert systems for large plant operation. In addition, Tsukuba University has produced some research work regarding the application of Genetic Algorithms (GA) to achieve optimality in the results of an expert system. The Toko Seiki Company and Osaka University have jointly produced some research work regarding the creation and development of methodology for reusable knowledge bases that may further be used in a variety of expert systems. (Kun, Guangyao and Lina) have done research to overcome the short comings of expert system such as weal knowledge acquisition and empirical knowledge acquisition uncertainty. They have presented a strategy on fault diagnosis expert system whose development is based upon fuzzy theory and ANN (ANN:artificial neural network) technique. Using fuzzy logic the fault symptom vectors are quantized, which are then inputted into the ANN to ratiocination, for the indication of the fault orientation. It is believed that the integration of these two techniques can improve the expert system’s performance when used for the fault diagnosis of the diesel fuel system. Fault diagnosis expert system based on this idea has been practically implemented. (Zhu, Zhang and Sun) have proposed the use of data mining and web technologies for GIS based agricultural expert system. The main idea is to use GIS and ES in combination as GIS can manage the acquisition of the geography information that is supposed to be input of the ES decision process. It can improve the precision and effectiveness of the ES decision by using the analytical result of GIS. Using network technology the decisions can be sent to the farmer, sooner and more conveniently. (Uzoka) presented a framework for the cost benefit analysis of an enterprise’s information system, using fuzzy ES. Fuzzy logic is intended to include unstructured variables such as the rating of the confidence of domain experts and decision makers in expert system’s decision, while performing the cost benefit analysis of an enterprise’s expert system. (Pidgeon and Freeman) have researched on the issues regarding the development of an ES in the domain of software design. The idea is to use the capabilities of an expert system for high quality, low cost software design. This research basically focuses on the automation of tasks of human designer and presents architecture to support the automation. (Ding) has produced research work related to the implementation of an expert system that may use AI techniques and networking technologies such as internet, to provide service for expert-level fault diagnosis for remote devices. Author has presented architecture for an expert system that is supposed to support the evaluation and testing of remote devices and may also support fault analysis and diagnosis in accordance with the test data, providing users with expert-level diagnosis services. Generic Design of an Expert system According to (Abraham) and (Choi) an expert system is computer software that imitates the process of thinking of a human expert in order to solve complicated problems in a particular domain. An expert system works as a system that interacts with the user, gives response to user queries, asks for explanation, gives suggestions, and generally aids the decision-making process. Expert systems serve to provide an expert’s guidance in a wide variety of activities, from computer diagnosis to medical and agriculture. An expert system usually comes up with three components: a knowledge base, an inference engine and a user interface. Figure 1 illustrates the generic architecture of an expert system: Figure 1: Architecture of Expert System Image Source: http://www.ucgis.org/summer2002/choi/choi.htm The knowledge base is the storage place for all appropriate domain information, data, rules, cases, and relationships that are used by the expert system to make decisions about any given query. It is allowed to combine the knowledge of more than one human expert in a knowledge base which enhances the precision of the decision made by an expert system. Usually knowledge about any domain is presented in form of conditional rules. A rule is a conditional statement that defines actions to be taken in response to given conditions. Knowledge base usually contains two different types of knowledge that is acquired from a human expert. It consists of facts and heuristic information. Facts are the known data about the problem domain and the structured or unstructured variables that may define any given activity. Heuristics contains the if-then logic that would be used by a human expert while making inferences. Knowledge is acquired by following a formal process that may include identification, formalization and conceptualization after which expert databases (knowledge bases) are developed (Abraham) and (Choi). The main aim of the inference engine is to explore the information in knowledge base and look for relationships between different facts in order to provide answers, advices, and suggestions by manipulating the knowledge base in the same way as a human expert. The accuracy of an expert system’s decision depends upon the capability of inference engine in finding the right facts, interpretations, and rules and assembling them correctly. Inference engines usually adopt two types of inference methods. First is backward chaining that is the process of starting with expected decision or goal and working backward to find the evidence for the goal and the supporting facts. Backward chaining is usually applied when expected goal is known and number of expected outcomes is not large. Second method of inference is forward chaining starts with the facts and works forward to reach the goal (Abraham) and (Choi). The user interface component contributes for the presentation part of the expert system and is used for designing, updating, and using expert systems. The user interfaces are designed for making the expert system easy to use for expert system’s developers, users, and administrators. It allows the user to query the expert system and receive the results (Abraham) and (Choi). The purpose of explanation facility is to allow a user to be aware of how the expert system reached at a certain decision. It enable the user to ask questions about “what”, “how”, and “why” aspects of a problem. The expert system will then provide a trace of the path to the user that lead the expert system towards the conclusion process, highlighting the key reasoning facts used during the process (Abraham) and (Choi). The knowledge acquisition facility is intended for the maintenance of the knowledge base in order to provide a suitable and resourceful means for capturing and storing domain information and heuristics in the knowledge base (Abraham) and (Choi). In addition to typical rule based ES, fuzzy ES’s have been developed in order to model the real word uncertainties, the human way of thinking, reasoning and perception processes. A fuzzy ES uses fuzzy logic and a collection of membership functions and rules that are used to reason about data instead of Boolean logic. Fuzzy ES arrives at the conclusion by performing a kind of numerical processing (set theory) unlike conventional ES that usually involve symbolic reasoning engines (Abraham) and (Choi). Applications of Expert systems Since the development of an ES in not a newer idea, with the passage of time, several industries have adopted this technology in order to get an expert’s advice in respective domain. The range of applications of ES technology to resolve the industrial and commercial problems is so broad so it cannot be described easily. The ES has its applications into most areas of knowledge work and has been developed for several industries and problem areas such as diagnosis and trouble shooting of all kind of systems and devices, planning and scheduling, financial decision making, knowledge publishing for delivering the knowledge that is relevant to user’s problem, process monitoring and control for detecting anomalies and predicting trends and design and manufacturing (WTEC Hyper-librarian) and (Turban, Leidner and McLean). (Wai, B and Rahman) describe the application of ES in the domain of agriculture. The ES for Agriculture provides the same functionality as others knowledge based system, it uses the rule base which is the information and knowledge of a human expert and it is represented in the form of condition- action rules and facts. This knowledge base is used by ES to solve problems by responding to the queries typed through a keyboard by a farmer on a variety of topics such as pest control, the need to spray, selection of a chemical, mixing and performance of spray, optimal management of machinery, recovery from weather damages such as freeze, frost or fog etc. in addition this research highlights various ES’s that have been developed to serve agricultural domain, including: National Institute of Agricultural Extension Management (MANAGE) has managed to develop an expert system called Rice-crop doctor that makes a diagnosis about expected attack of pests and diseases for rice crop and proposes defensive or healing measures. The rice crop doctor provide its service in the area of rice production through development of a prototype, by constructing a knowledge base about a few major pests, diseases and some problems limiting the production of rice (Wai, B and Rahman). An Expert System called AGREX have been prepared by Center for Informatics Research and Advancement, Kerala to help the farmers by giving them timely and correct advice about their crops. This system covers consultancy for a broad range of issues including the fertilizer’s application, crop protection, irrigation scheduling, and diagnosis of diseases in paddy and post harvest technology of fruits and vegetables (Wai, B and Rahman). Similarly, VARIEX is an ES that has been prepared at Technical University of Brno, Czechoslovakia which supports the selection of the best cultivators for different agricultural conditions (Wai, B and Rahman). Site-Specific Technologies for Agriculture (SST4Ag) have developed an ES called DSS4Ag that uses state-of-the-art AI techniques and computer science technologies to make geographical area based site-specific, best possible decisions for the use of fertilizer (Wai, B and Rahman). (Khan, Maqbool and Razzaq) describe the application of ES in medical domain. Medical ES’s provide support for decision making in certain situations where either the problem is quite complicated or there is a problem of instant availability of medical experts for patients. ES are expected to play a great role in medical domain by providing expert’s opinion in common clinical problems like diagnosis and prevention of diseases and counseling of the patients etc. (Grove) identifies internet based ES’s that have been developed to serve the medical domain. HEPAXPERT/WWW is an Internet-based interface to an ES that provide support for the interpretation of serologic tests for hepatitis infections. Queries are entered by user through a Web portal, after off-line processing conclusions are sent to user via e-mail. Similarly, The Protocol Assistant is an Internet-based ES facilitating the diagnosis of parotid tumors. (Jurgen Dorn et al., 1996) describe the application of an ES in steel industry. Steelmakers find it convenient to apply ES’s instead of conventional software because the reasoning power of controlling ES with existing uncertainties tends to lessen the inherent complexity of the respective decision making process. The ES in steel industry is supposed to address two problems. Firstly, the prediction of irregular situations such as abnormal and unexpected shortage of the raw materials charged in the heating system and the detection if the heated gas reaches the top of the furnace without reaction. Secondly, the ES takes care of keeping the thermal condition stable. Conclusion ES’s provide a convenient medium for providing expert’s advice in a wide range of domains and play a major role in solving the problems such as inability to meet the domain experts in decision making process and to get the proper understanding of respective domain. The development of an ES is not a newer idea but the blend of AI techniques, computing technologies and networking technologies have enabled the development of such an intelligent ES that may enable user to accurate expert’s opinion anywhere at remote locations. Despite of aforementioned advancements, the generic architecture of an ES almost the same and consist of a knowledge base, an inference engine and a user interface. Additionally, the explanation facility and knowledge base acquisition facility can be incorporated in the ES’s architecture for the formalization of knowledge acquisition and decision making process. ES’s have their application in multiple domains, replacing the role of domain experts, that is quite helpful for the people in those domain for solving the problems that require the opinion of a domain expert. Various ES’s have been developed to serve the agricultural domain with the aim of diagnosing the diseases, predicting possible pest attacks, suggesting chemicals to spray, appropriate time for spray, appropriate time for fertilizer etc. ES’s are also providing services in medical domain regarding the diagnosis of decease, suggestion of medicine and counseling of patients etc. Similarly, ES’s have also been applied in the steel industry, providing solution for various domain specific problems. Abbreviations ES Expert System AI Artificial Intelligence GA Genetic Algorithms IKA Integration of Knowledge Ability Work Cited Abraham, Ajith. Rule- Based Expert Systems, Handbook of Measuring System Design. New York: John Wiley & Sons, 2005. Choi, Jinmu. "A Rule-Based Expert System Using an Interactive Question-and-Answer Sequence." 2010. 25 December 2010 . Ding, Zhiqin. "Research on Internet based Open Remote Fault Diagnosis Expert System." International Conference on Information and Multimedia Technology (ICIMT) 2009. Baku : IEEE, 2009. 1-4. Grove, Ralph. "Internet-Based Expert Systems." Expert Systems Volume 17, Issue 3 (2000). Hou, Guangming and Jianmin Fan. "Research on Virtual Expert System and Its Construction Based on Integration of Knowledge Ability." Business and Information Management, 2008. ISBIM '08. International Seminar. Wuhan : IEEE, 2008. 240-243. Hutchinson, Sarah E. and Stacey C. Sawyer. Computers, Communications, Information A user's introduction, 7th Edition. New York: Irwin/McGraw-Hill, 2000. Khan, Fahad Shahbaz, et al. "The Role of Medical Expert Systems in Pakistan." World Academy of Science, Engineering and Technology (2008): 280-282. Kun, Yang, OuYang Guangyao and Ye Lina. "Research upon Fault Diagnosis Expert System Based on Fuzzy Neural Network." 2009 WASE International Conference on Information Engineering. Taiyuan, Shanxi, China : IEEE, 2009. 410-413. Laudon, Kenneth. C. and Jane. P. Laudon. Management Information Systems, Sixth Edition. New Jersey: Prentice Hall , 1999. Mizoguchi, Riichiro and Hiroshi Motoda. "Expert Systems Research in Japan." Intelligent Systems Volume 10, Issue 4 (1995): 14-23. Pidgeon, Christopher W. and Peter A. Freeman. "Development Concerns for a Software Design Quality Expert System." Design Automation, 1985. 22nd Conference. ACM, 1985. Turban, Efraim, et al. Information Technology for Management: Transforming Organizations in the Digital Economy . New York: Wiley, 2005. Uzoka, Micheal E. " Fuzzy- Expert System for Cost Benefit Analysis of Enterprise Information System: A Framework." International Journal on Computer Science and Engineering Volume 1, Issue 3 (2009): 254-262. Wai, Kiong Siew, et al. "Expert System in Real World Applications." 2007. Generation5.com. 25 December 2010 . WTEC Hyper-librarian. "The Applications of Expert Systems ." 1993. 26 December 2010 . Zhu, Zhiqing, Rongmei Zhang and Jieli Sun. "Research on GIS-based Agriculture Expert System." Software Engineering, 2009. WCSE '09. WRI World Congress. Xiamen: IEEE, 2009. 252-255. Read More