2.2 Knowledge Acquisition
Knowledge acquisition is a process which aims at extracting knowledge, experience and problem- solving procedures from one or more domain experts. There are many general and specific knowledge acquisition approaches available. A general knowledge acquisition approach consists of five stages as shown in Figure 1. The first stage, elicitation, aims at extract- ing domain experts’ knowledge about the prob- lem under consideration. Modelling is the pro- cess for transforming the findings of the first stage into more formal knowledge representa- tions. There is a feedback loop between these two stages that allows to make corrections of in- consistencies which might have caused by misun- derstanding between the expert and the knowl- edge engineer. Internal testing is verification process to ensure the correctness of the develop- ment before field trials can be proceeded. The field trial requires the involvement of all relevant parties to validate the developed product.
80
The efficiency and effectiveness of the process of knowledge acquisition are affected by the fol- lowing [5]:
Experts may not be able to express their knowledge and experience in a structured way.
Experts may not be aware of the signifi- cance of the knowledge they have used.
Even experts can express their knowledge. it may be irrelevant, incomplete, incorrect and not understandable
Normally. domain expert are not necessary to fa- miliar with computers. In order to acquire their knowledge to meet the requirements of the devel- opment of expert systems. knowledge engineers are needed to bridge the gap between a computer system and domain experts. There are many methods used in knowledge acquisition and could be classified into 2 groups [lo], namely direct and indirect methods.
e Direct Approach: The knowledge engineer asks the expert to report on knowledge that he can directly articulate. The following methods are commonly used as direct ap- proach.
Interview Questionnaires - Observation of the task performance - Protocol analysis - Interruption analysis Drawing closed curves - Inference flow analysis
e Indirect Approach: The expert are not asked to express their knowledge directly but are asked some questions. By ask- ing these questions, other behaviour is col- lected. Some examples of indirect approach are as following:
- Multidimensional scaling - Johnson hierarchical clustering - Ordered tree from recall - Repertory grid analysis
The choice of the methods to be used depends on the domain field, the domain expert and the real sit>uation. It is not uncommon that more than one method are used to ensure full understand- ing of the real life situation. However, interview- ing the domain expert is frequently included as part of a knowledge acquisition process. Normally the knowledge acquisition is an iter- ative process which allows formulated knowledge to be verified and validated from time to time. In brief, the knowledge engineer extract knowl- edge from domain expert and then organise it in the form that can be utilised by an expert system. If the expert system can not provide satisfactory outcome, the domain expert will be consulted again for improvement. The amount of effort expended and the number of iterations required depend on the size of the system to be built, the depth and breath of the tasks to be supported, and the quality of the knowledge as it is acquired [9].
2.2 Knowledge Acquisition Knowledge acquisition is a process which aims at extracting knowledge, experience and problem- solving procedures from one or more domain experts. There are many general and specific knowledge acquisition approaches available. A general knowledge acquisition approach consists of five stages as shown in Figure 1. The first stage, elicitation, aims at extract- ing domain experts’ knowledge about the prob- lem under consideration. Modelling is the pro- cess for transforming the findings of the first stage into more formal knowledge representa- tions. There is a feedback loop between these two stages that allows to make corrections of in- consistencies which might have caused by misun- derstanding between the expert and the knowl- edge engineer. Internal testing is verification process to ensure the correctness of the develop- ment before field trials can be proceeded. The field trial requires the involvement of all relevant parties to validate the developed product. 80 The efficiency and effectiveness of the process of knowledge acquisition are affected by the fol- lowing [5]: Experts may not be able to express their knowledge and experience in a structured way. Experts may not be aware of the signifi- cance of the knowledge they have used. Even experts can express their knowledge. it may be irrelevant, incomplete, incorrect and not understandable โดยปกติ ผู้เชี่ยวชาญโดเมนไม่จำเป็นต้อง miliar fa กับคอมพิวเตอร์ เพื่อรับความรู้เพื่อตอบสนองความต้องการของ opment devel ของระบบผู้เชี่ยวชาญ วิศวกรความรู้จะต้องเชื่อมต่อช่องว่างระหว่างตัวคอมพิวเตอร์โดเมนและระบบผู้เชี่ยวชาญ มีหลายวิธีใช้ซื้อความรู้ และสามารถจำแนกได้เป็น 2 กลุ่ม [หล่อ], คือวิธีทางตรง และทางอ้อม อีวิธีการโดยตรง: วิศวกรความรู้ถามผู้เชี่ยวชาญรายงานความรู้ที่ได้โดยตรงสามารถบอก โดยทั่วไปใช้วิธีต่อไปนี้เป็น ap proach โดยตรง สัมภาษณ์โพรโทคอลแบบสอบถาม -เก็บข้อมูลประสิทธิภาพการทำงานของงาน - วิเคราะห์ -เส้นโค้งวาดปิดหยุดชะงักวิเคราะห์ - ข้อวิเคราะห์การไหล อีวิธีทางอ้อม: ผู้เชี่ยวชาญขอไม่แสดงความรู้โดยตรง แต่จะถามคำถาม โดยถาม-ing เหล่านี้ถาม พฤติกรรมอื่น ๆ เป็นคอลัมน์ lected ตัวอย่างของวิธีการทางอ้อมมีดังต่อไปนี้: -มิติมาตราส่วน - Johnson ลำดับคลัสเตอร์ - สั่งต้นไม้จากการเรียกคืน - ละครตารางวิเคราะห์ The choice of the methods to be used depends on the domain field, the domain expert and the real sit>uation. It is not uncommon that more than one method are used to ensure full understand- ing of the real life situation. However, interview- ing the domain expert is frequently included as part of a knowledge acquisition process. Normally the knowledge acquisition is an iter- ative process which allows formulated knowledge to be verified and validated from time to time. In brief, the knowledge engineer extract knowl- edge from domain expert and then organise it in the form that can be utilised by an expert system. If the expert system can not provide satisfactory outcome, the domain expert will be consulted again for improvement. The amount of effort expended and the number of iterations required depend on the size of the system to be built, the depth and breath of the tasks to be supported, and the quality of the knowledge as it is acquired [9].
การแปล กรุณารอสักครู่..