An Ontology-based Approach to Plant Disease Identification System

Disease identification in plants is an important issue for farmers in terms of plant production and the reduction of losses in crop field. To deal with this issue, a number of systems and techniques for identifying plant diseases have been proposed. However, most of them mainly concentrate on image-based pattern recognition rather than focusing on the observed abnormalities of plant diseases. In other words, they have not employed ontology for semantic detection of plant disease. Current systems do not support farmers to find disease names w.r.t. the semantics of an infected plant. In this work, we proposed an ontology-based approach to modeling plant diseases and demonstrate our approach by developing a rice disease ontology. The ontology helps identifying plant diseases from existing symptoms on plants. To construct the ontology, we reused the domain knowledge related to symptoms of plant diseases from reliable sources. The proposed ontology is modeled in Web Ontology Language (OWL). We also develope a system architecture compatible with the modeled ontology. Finally, we illustrate the usage of our ontology in DL Query for disease retrieval, given that a farmer's observation has already been transformed into an OWL concept. The retrieved results of DL Query make use of subsumption relationship among concepts and properties defined in the ontology.