An Ontology-Based Approach for Diagnosing Date Palm Diseases

  • Mahmoud A. El-Askary -----> Dr. Rebhi S. Baraka

Date Palm is one of the oldest fruit trees in the world and is deeply rooted in the economics, history and culture in the Arab world. Because of its economic and social importance, date palm has a high research priority for further development of crop production and protection using the best approaches that modern science and technology can provide. Date palm trees as the rest of the fruit trees are exposed during their growth to many different pests that cause high economic damage to production. There are symptoms that appear on the plant which must be diagnosed quickly to make the right decision as a prevention. Pest control methods are the processes that lead to the reduction of pest's damage to plants by limiting the spread and reproduction.
In this research, we propose an approach that aids the development of a plant protection expert system for date palm. It is based on the ontology concept to diagnose the disease and suggest appropriate treatment by identifying anomalous observations on the parts of the tree. The approach consists of three inter-related components:
knowledge base, reasoning engine and server side application. The knowledge base is built using OWL ontology and contains knowledge about date palm diseases and insectpests,
named for AgriDPalmOnto. The reasoning engine accepts user input queries and responses to data through the I/O interface and uses this dynamic information together with the static knowledge stored in the knowledge base. The web application works as an interface to the system where the user enters his queries and gets system feedback and answer. We evaluate the approach according to a human expert in plant diseases by comparing his diseases diagnoses to those of the system, system showed good accuracy in the results were 83.5% compared to documented scientific answers. The result is better than the agricultural expert's. We evaluate the ontology using Task-Based framework it indicate that the accuracy of using the AgriDPalmOnto is 100% and 96.7% when using evaluation method precision and recall. In addition, we use SPARQL queries to insure correct feedback from ontology.