Knowledge Management In Small and Medium-Sized …



Knowledge Management In Small and Medium-Sized Enterprises (SMEs): A Case Based Reasoning Approach

GEORGE NATHANIEL PAPAGEORGIOU & GOPI BOMMAREDDY

Department of Management Information Systems

Cyprus College

Nicosia

CYPRUS

ABSTRACT: -- Knowledge plays an important role in each and every organization. Acquiring, maintaining and managing knowledge in a meaningful way for future use could be a source of sustainable competitive advantage for any organization. This paper describes a new Case Based Reasoning (CBR) approach in order to manage knowledge in small and medium-sized enterprises. CBR, which is an Artificial Intelligence (AI) technique, stores the previous experiences as cases in a case base and retrieves the suitable cases from the case base according to a present problem situation. Major tasks in the CBR are Retrieve, Reuse, Retain, and Revise past relevant cases. This paper contains the main facets of CBR, its advantages and disadvantages, in relation to other methods, which are available to develop knowledge-based systems. After extensive literature review in this paper we propose a new framework to develop a CBR system. The new proposed framework takes a holistic approach in developing a knowledge-based system by examining people, organizational and technical issues. The framework developed will be useful for the management of small to medium sized organizations, consultants, and developers who would be interested in developing knowledge management systems.

Key words: -- Knowledge Management, Managing Knowledge in Organizations, Case Based Reasoning, Small and Medium based Enterprises, Knowledge Retrieving, Knowledge Representation.

1.Introduction

1.1 Knowledge Management

Knowledge Management is the process of acquiring knowledge from the intellectual and knowledge based assets of an organization. Generating value from such assets involves sharing them between the employees, departments and with other organizations. Even though we acknowledge the fact that knowledge is something that resides in people’s minds rather than in computers [24], it is of outmost importance for organizations to store a knowledge base in a computerized system that can be shared among organisational members.

We can consider knowledge to be of two types Tacit and Explicit.

Explicit Knowledge: -- Knowledge, which we can share through the information technology. We can store this knowledge in documents, manuals, computers etc. This knowledge is very easy to retrieve in future [8].

Tacit Knowledge: -- It is also called as implicit knowledge. This is the knowledge, which we cannot see or we cannot store manually. It is in the expert’s minds. This knowledge is related to the skills of the experts [8].

It is important for every organization to store both types of knowledge for future usage. Explicit knowledge is easy to store and retrieve. But tacit knowledge, which is in the expert’s mind, is difficult. And this knowledge is very important for an organization. For example if an expert leaves an organization a significant amount of knowledge is lost. But if we have a system to store the tacit knowledge, this remains with the organization. Knowledge based systems are introduced to manage the explicit and tacit knowledge in organizations. Using a variety of AI techniques these systems can be developed.

1.2 Case Based Reasoning

“Case Based Reasoning (CBR) is an artificial intelligence approach to solve a new problem by remembering a previous similar situation and by reusing information and knowledge of that situation” [27]. CBR is a very effective methodology to solve problems “Where the acquisition of the case-base and the determination of the features is straight forward compared with the task of developing the reasoning mechanism” [5]. In order to solve a new problem, CBR may adapt and combine old solutions and explain new situations in accordance with the old similar problems. All these different aspects are classified into two major categories; they are Interpretative CBR system, and problem solving system [18]. Based on the similarities and differences among the new case and past case the interpretative system will propose the solution to the present problem. Problem solving system proposes the solution by adapting the previous cases [21][23].

In this paper we present a framework that utilizes both CBR types in order to develop knowledge management systems in small to medium enterprises. Most of the CBRS performance depends on the maintenance of cases in the case base, case retrieval, understanding and adaptation of the new cases into the case base. Based on retrieving the similar cases from case base, selecting the suitable cases from the retrieved cases, deriving a solution from best cases, evaluating the proposed solution, storing the new case into case base, Aamodt and Plaza describe the CBRS as a cyclic process which contains the four R’s that is Retrieve, Reuse, Revise and Retain [1][3][23].

1.2.1 History of CBR

Work of Schank and Abelson in 1997 was the basic foundation to the CBR research. They proposed a framework to store the general knowledge as scripts, which store the expectations. The disadvantage in this system is that it creates the confusion with most similar scripts. Due to this they concentrate more on format and structure of the scripts, problem solving method, learning and philosophy and psychology. At the same time Gentner developed a framework for analogy, which was related to CBR. Finally Kolodner developed the first CBR application CYRUS. It was developed on the basis of the Schank’s dynamic memory model. In this system it stores the details of the travels and meetings of the ex-US secretary –of- state Cyrus vance. Later MEDIATOR, CHEF a meal planning system, PERSUADER, CASEY, and JULIA were developed based on the CYRUS system [23][30].

In Europe the first CBR research work started in Scotland. Derek Sleeman’s group started work by studying the use of cases for knowledge, and they developed the REFINER system, which is used to refine an expert knowledge in a more natural way. CBR research is not only limited to the United States and Europe but also expanded to many other countries across the world such as India, Japan, and Israel. CBR has a diverse range of applications, which varied from medical diagnosis, planning, help desks, problem solving and architecture design [23][29].

Based on the above facts in recent years CBR is one of the most successful Artificial Intelligence (AI) approaches in building knowledge-based systems. It does not depend on rules and relationships. But instead, it depends on the previous cases and experiences [27]. The basic idea of the CBR systems is to remember past problems and situations to reuse in the future. Thus CBR became different approach than other approaches. In order to solve a new problem, CBR may adapt and combine old solutions and explain new situations in accordance with the old similar problems. And CBRS always collects the knowledge and experiences of the experts who are not available every time for consultation and store it in the form of cases in the case base [26].

1.2.2 Advantages of CBR

• A CBRS can be developed with a relatively small knowledge base. It does not require complete knowledge of the problems and the situations [27].

• Through a CBRS we can learn easily, by simply adding a new case to the case base by adaptation of the case [13].

• CBR reduces the time of problem solving by allowing the user to propose the solutions to the problems quickly. And there is no need to understand the complete problem [3][4].

• No need to propose the new solutions for every case from scratch. It uses the previous experiences.

• It allows the user to evaluate the case without using any specific algorithmic method.

• It also allows the incomplete, open ended and ill-defined cases to store in the case base [19].

• If a small case base is available for a CBRS, then it will start reasoning with the small case base. It expands the case base by adding the new cases.

• Users can be trained fairly quickly as to have to add a new case or how to adopt cases.

1.2.3 Disadvantages of CBR

At present there is no common framework to implement a CBR system. Such a framework would be useful for understanding the state of the art in CBR maintenance, illuminating current practice and facilitating the comparison of particular approach as has already proven useful for studying CBR [17][20].

• Most of the current CBR systems are based on a single case base. By using similarity checks and problem description it can retrieve the most suitable cases.

• Maintenance is the major problem with the single case base system. It is very difficult to retrieve the similar cases, difficult to allocate the indices to each case for future retrieval.

• Making each case distinct and relating it with other similar cases is difficult.

• A case base consists of too many cases. If it is a single case base, then it will take more time to retrieve the suitable cases. So it is a time consuming process.

• The defect in this system can also be seen in the performance and accuracy of the system. This means an existence of flaw in retrieving the most suitable cases at desired time.

• It provokes a user to apply the past cases to present situation with out any prior checking.

2. The Proposed Framework

There are many methodologies to develop Information Systems (IS), but no methodology can be universally applicable to all organizations. So rather a situational based method should be adopted for each case [25]. Therefore for the case of developing knowledge management systems for small to medium sized enterprises we propose the following framework.

As seen in figure 2, we suggest a cyclic mode with the use of prototypes in examining the current situation, identifying tasks and issues, specify requirements in order to design the knowledge management system, which is based on the case based reasoning. Our framework makes use of concepts form Soft Systems Methodology and Rapid Application Development. The aim is to use our framework to develop a knowledge management system that will be flexible to maintained and easy to use as a tool of sharing knowledge.

Knowledge Management system in every organization must be easily identifies the knowledge in the organization, flexible to managing, sharing and using that knowledge in proper way. The most important aspect is using the best method to retrieving and reusing that knowledge in appropriate situation. Thus we propose a new CBR approach to build the knowledge management system an organization.

2.1 CBR in the Proposed Framework

Most of the CBRS depends on the quality and the quantity of the knowledge in the case bases. The possibility to solve the problem becomes limited when the knowledge is less and problem solving of CBRS decreases. Learning from the knowledge, which is gathered, is a continuous process in CBRS. Learning is a primitive art for human beings, but undertaking the job of building a system is a complicated one. As seen in figure 3, the design of the CBRS concentrates on case representation, indexing, and retrieving.

2.2 Case Representation

In the proposed framework we divide the single case base into different case bases based on the category of the case. The categories are formed based on the features of each case. Before we add a new case into the case base we check the features of the case and then we add it to the relevant case base. This type of representation forms a hierarchical structure of the case base. In general every case structure is combined with a problem description, solution, and outcome of the case. In addition to problem description, solution, and outcome of the case, we propose addition of a couple of new sections to each case i.e. Connection and Keywords. Connection section stores the indices of the most related cases. Using this section it is very easy to build the connection between the cases and different categories. The usage of these relationships results in the decrease of CBRS retrieval time and an increase in the problem-solving efficiency. It reduces the complexity of the CBR. While retrieving the cases, the CBRS retrieves the related cases based in the similarity checking with the help of this connection section too. This method creates the chain type relation between the different cases and case bases. In the keyword section we store the most related keywords to that case which it facilitates the user for different types of searching.

2.3 Case Indexing

Indexing of cases is one of the major tasks in the CBR. Retrieving of the cases depends on the indexing. Indexing is also two types 1) automatic indexing 2) manual indexing. CBRS allocate the index for the new case based on the pre determined rules. Before we add the new case, we must prepare the rules for automatic indexing based on the features and content of the cases. The user can allocate the index when new case is added to the case base. Depending on the situation user can choose any type of indexing

2.4 Case Retrieving

2.4.1 The use of Index & Keywords

The proposed CBR framework allows the user to retrieve the case directly from case base by using the index of the case. This method doesn’t take much time to retrieve the cases. It reduces complexity and process time, but at the same time it is difficult for the user to remember all indices of the cases. Hence, we propose that this method has to be combined with searching. Users can search the case base by using the related keywords, which are stored in the keywords section of the case.

2.4.2 The use of Induction and Nearest Neighbour Algorithm

The retrieving process of the proposed framework depends entirely on the use of both the induction and nearest neighbour algorithm. Most of the CBRS applications use either the nearest neighbour or induction method for retrieving the related cases. In this method, we calculate the similarity between the past cases and the present situation. If the case base is small, then one of the above methods works effectively. If case base become large, then complexity of retrieving the cases also increases. This increase is due to the increase in the number of cases searched. To overcome this problem, we propose a framework with the combination of the induction and nearest neighbour algorithm for retrieving and selecting the most relevant cases. In the proposed framework, first we use the induction approach to retrieve the related cases and create the hierarchical structure of the related cases from the case base. After creating the hierarchical structure the nearest neighbour algorithm selects the most related cases from that structure based on the features of the case.

Induction Approach: -- This collects all the related cases from the case base and forms the decision tree type structure for selecting the most related cases from the structure. It creates the decision tree based on the problem description, solution, and features of the cases. This induction method is very useful when only one feature as the solution for the present problem is required [31].

Nearest Neighbor Algorithm: --

Where W i is the important weight of a feature, Sim is the similarity function of features, and f iI and f iI are the values for feature i in the input and retrieved cases respectively

3. Application of Proposed Framework

The proposed framework was applied in order to build a Knowledge Management system in a medium size hardware sales and servicing company in Nicosia, Cyprus.

3.1 About the current system

Figure 4 provides a pictorial representation of what the current problem situation is in our hardware sales and servicing company. From this pictorial representation (Rich Picture) one can identify major tasks and issues, which need to be addressed to improve the current situation. The current system in the technical department is fully clerical. They are not using any computer-based system to store and retrieve the best cases for future use. Simply they store the manuals of each machine. The technicians repair the machines using these manuals. If the technician is unaware of the solution, then he searches for the related manual. Upon getting the solution he repairs the machine. And if he doesn’t get any information about that particular problem from the manuals then he asks for help of the technical manager. If both manager and technician don’t manage to solve the problem, then they ask for help from the manufacturer company of that machine and then they update the manuals for future use.

But obviously the above is a very tedious, time consuming process. Further if any technician leaves the company he takes all the acquired knowledge with him, which is in his mind. So most of the knowledge is in tacit form and is related to the skills and experiences of the specific technician. If any technician needs support while examining a particular problem, has to go back to the office to review the manuals. That of course a causes delays in the service and negatively affects customer satisfaction. Therefore it is imperative to develop a knowledge management system where employees can share their experiences and there by solve problems quickly and accurately.

3.4 Development of the CBR system

By going through several prototypes as described in the proposed framework, we have developed a knowledge management system for our hardware sales and servicing company, which we have named Case Based Technical Support System (CBTSS). What it follows is a description of how the CBTSS works.

Figure 5, shows the CBTSS user interface, which has variety options. Problem description, features, indices list, solution, buttons to select the different type of searching, buttons to adding and modifying the cases. The inputs are entered in the problem description area. The CBTSS enables search by using the options such as problem, feature, index and, keywords. By choosing the problem button, we can see the list of related case indices in list the box. The user selects the desired case index. The indices in the connection section of the case are displayed in the related case list. As shown in Figure 5 by selecting either case index or related case index, the information about the case is generated. Further search can be carried out by problem and by feature in order to achieve refinement of the results and generate the most related cases. Concerning the feature the use of nearest neighbour algorithm is used in order to enable the case base search as seen in figure 6. Finally the search by keyword option, such as MEM, CPU, HDD, and RAM etc is demonstrated in figure 7.

4. Discussion & Conclusions

In this paper we have proposed a CBR approach to develop knowledge management systems for small to medium sized enterprises. In CBR the major concern is gathering knowledge from experts maintaining that knowledge and using it for future problems. The gathering of knowledge proves to be difficult due to the fact that is mainly in a tacit form. In some situations due to lack of sufficient cases and proper case representation retrieving of relevant cases becomes difficult. If too many cases are represented in a single case base, it leads to problems in maintaining cases, and retrieving suitable cases, which affects negatively the process of problem solving. In order to overcome these problems we suggest the use of multiple case bases in a hierarchical representation as seen in the proposed framework.

Although we have the above issues, CBR is a preeminent method to develop a knowledge-based system for an organization. Comparing with the other methods, CBR requires little knowledge acquisition. In most of the methods it is difficult to create rules to acquire and maintain the knowledge. But in CBR adding the new cases to the case base is not a complex task. To maintain a knowledge base an expert is required in other systems, but in CBR system a user can add, modify, and delete the cases. In CBR system users don’t need any programming or any special skills to maintain the case base. In a CBR system one more advantage is saving the failure cases in a distinct case base. By this way we can avoid the repeating the similar problems and increase the problem solving efficiency.

The future of CBR lies in being adopted by everyone on a commercial basis. It’s future also lies in getting adopted as an e-business solution in many organizations. We perceive that the proposed cased based reasoning framework can be extended to wider horizons of rule based and model based reasoning approaches. Since we have concentrated on representation, induction and retrieving, we suggest that further research can be carried out in the areas of reuse, retain and in adaptation of case.

References:

[1] Aamodt, A. & Plaza, E. (1994). Case-Based Reasoning: Foundation Issues: Methodological Variations, and System Approaches. AI Communications, pp. 39-59.

[2] Althoff, K.D. & Wess, S. (1992). Case-based reasoning and Expert System Development. In, Contemporary Knowledge Engineering and Cognition, Springer-Verlag, Lecture Notes in Computer Science 622, pp.146-160.

[3] Andrew, B. (1997). Case Based Reasoning Advanced Knowledge-BasedSystems.

[4] Arijit, S., Wilson, C., & Leake, D. (1999). Proceedings of the workshop on practical Case-Based Reasoning strategies for building and maintaining corporate memories. Third international conference on Case based Reasoning, Germany.

[5] Bonzano, A., & Cuningham, P. (1996). ISAC: A CBR System for Decision support in Air Traffic Control in Proceedings of EWCBR '96, Advances in Case-Based Reasoning, Ian Smith & Boi Faltings eds. Springer Verlag Lecture Notes in AI, pp. 44-57.

[6] Checkland, P. & Scholes, J. (1990) "Soft Systems Methodology In Action", John Wiley and Sons.

[7] Christos, S. & Dimitrios, S. (1996). Neural Networks. Surveys and Presentations in Information Systems Engineering. Vol 4.

[8] Duffy, J. (2000), Knowledge management: to be or not to be? Information Management Journal, Vol. 34 No. 1, pp. 64-67.

[9] Gurney, K. (1994). Computers and Symbols versus Nets and Neurons. Psychology Department, University of Sheffield.

[10] Hammond, K. (1989). On Functionally Motivated Vocabularies: An Apologia. Proceedings of the Second Workshop on Case-Based Reasoning. FL, USA.

[11] Hammond, K (1989). Case Based Planning; reviewing planning as a memory task, Chapter 3, Academic press

[12] Hanney, K., Keane, M.T., Smyth, B. & Cunningham, P. (1995). When do you need adaptation? a review of current practice. In Proceedings of AAAI Fall Symposium on Adaptation, pp 41-46.

[13] Hua, K. & Faltings, B. (1993). “Exploring Case-Based Building design - CADRE”, Artificial Intelligence for Engineering Design, Analysis, and Manufacture, Vol. 7, No. 2, pp. 135-143.

[14] James, F.H. (2003). Introduction to Rule-Based Systems. Rule-Based Systems and Identification Trees.

[15] Joan Garfield. (2002). The Challenge of Developing Statistical Reasoning. Journal of Statistics Education Vol 10, No 3.

[16] Johanyak, Z. C. (1998). Representing cases in a case-based FMEA system. 9th Daaam International Symposium, Technical University Cluj - Napoca, Cluj - Napoca, Romania

[17] Jurisica, I. (1993). Representation and management issues for case-based reasoningsystems, technical report, University of Toronto, Department of Computer Science, Toronto, Ontario M5S 1A4, Canada.

[18] Kolodner, J.L. (1993). Case-Based Reasoning. Morgan Kaufmann Publishers, Inc

[19] Kolodner.J.L & Leake, D. (1996). A tutorial introduction to case-based reasoning. In D. Leake, editor, Case-Based Reasoning:Experiences, lessons & Future Directions. American Association for Artificial Intelligence.

[20] Shink, K. & Shin, I. H. (2001). A case based approach using inductive indexing for corporate bond rating. Seoul, South Korea.

[21] Leake, D. (1996). CBR in Context: The Present and Future. In Case-Based Reasoning, Experiences, Lessons, & Future Directions. ed. D. Leake, 3-30. Menlo Park, Calif. MIT Press.

[22] Leake, D.B & Wilson, D.E. (1998). Advances in Case Based Reasoning, Proceedings of EWCBR-98, Springer-verlag, 1998.

[23] Lopez, B. & Plaza, E. (1997). Case Based Reasoning: An Overview. AI Communications, 21-29.

[24] Maria, M. (2000). A critical review of knowledge management as a management tool. Journal of Knowledge Management Vo 4 (3) pp 204 - 216

[25] Papageorgiou. G. N, (2005). Towards improving the quality management information systems development through the use of a situational-based framework. Journal of business and society. Nicosia, Cyprus

[26] Ross, S., Fang, L. & Hipel, K.W. (2002). “A Case-Based Reasoning System for Conflict Resolution: Design and Implementation”, Engineering Applications of Artificial Intelligence, Vol. 15, No. 3-4, pp. 369-383.

[27] Shin, K. & Han, I. (2001). A Case-based Approach Using Inductive Indexing for Corporate Bond Rating, Decision Support Systems, Vol.32, No.1, 2001, pp.41-52.

[28] Simoudis, E. (1992). Using case-based reasoning for customer technical support. IEEE Expert 7(5), 7-13.

[29] Watson, I. & Marir, F. (1995). Representing and indexing building refurbishment cases for multiple retrieval of adaptable pieces of cases. In, Case-Based Reasoning Research and Development, Veloso, M., & Aamodt, A. (Eds.), Lecture Notes in Artificial Intelligence 1010, Springer-Verlag, Berlin.

[30] Watson, I. & Marir, F. (1994). Case based Reasoning: A Review, AI-CBR, The knowledge Engineering Review, 9-4, pp. 327-354.

[31] Watson, I. (1997). Applying Case-Based Reasoning: Techniques for Enterprise Systems. Morgan Kaufmann Publishers Inc.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download