DEVELOPING WEB BASED SELF-LEARNING EXPERT SYSTEM FOR DIAGONOSTIC ADVISORY OF SESAME DISEASES IN ETHIOPIA

Sesame is a strategically important crop to Ethiopia, as it ranks as a top exportable crop and plays a great role in the country’s economy. In the last few years (2007-2011), Ethiopia was the world's 3rd largest sesame exporter, supplying to China and Turkey, however, sesame production has been declining, though; there is still high potential for increased production. This is because of the shortage of professionals to diagnosis and controls the sesame diseases at the right time. So it is important to develop an expert system that helps the professionals as well as for the investors, farmers, and Ethiopians as a whole by diagnosis and giving advisory for the sesame diseases.
The objective of this study is to develop a web Based self-learning expert system for Diagnosing and Advisory of Sesame diseases. To develop for the prototype, the knowledge was acquired using both structured and unstructured interviews from selected sample domain experts and represented by production rule, and the Decision tree was used for modeling the knowledge of the sesame diseases and symptoms.
Several studies have been conducted using different approaches but they did not incorporate the self-learning and user-friendly interface. So in this study, we proposed a web-based self-learning expert system for the advisory of sesame diseases. Moreover, this expert system is developed using CLIPS(C Language Integrated Production Systems) for developing the knowledge base and ASP.NET as the programming language. The self-learning Web-Based expert system can give its diagnosis result and advisory system based on user preferences. And it is self-learning and user-friendly in which many users can easily interact with the web-based interface for diagnosis using a fast RETE pattern matching algorithm.
The developed prototype performance was evaluated using user acceptance testing and was evaluated and it scores 86.8 % according to the system evaluators. Besides the accuracy of the expert system was evaluated at 93.3%. In conclusion, developing the prototype in local languages, and applying other Machine learning algorithms are the future works of the study.