Diagnosa Hama dan Penyakit Cabai dengan Menggunakan Bayesian Network Berbasis Graf Pengetahuan

Abstract

Chili plants are seasonal plants which are included in the category of plants that are vulnerable to pest and disease attacks. One feature that has been successfully developed by several experts or researchers is an independent diagnosis system. However, the data structure of each system is certainly different, which can make it difficult to develop the system. A system is needed that can be used as an independent diagnostic system which is developed through integrated and structured data. This system was built using a graph-based Bayesian Network (BN) method which combines knowledge graphs, ontology and Bayesian Network to support the diagnostic process. Knowledge graphs are used to represent relationships between entities related to chili plant diagnostics, including symptoms, diseases and pests. Ontologies are used to provide a clear and defined knowledge structure framework for diagnostic entities. Bayesian Network is then built based on knowledge graphs and ontology to model probabilistic relationships between symptoms, diseases and pests in chili plants. The research method used in this research includes five research stages, namely: (1) definition of the problem and specification of objectives; (2) ontology development; (3) Bayesian Network inference; (4) demonstration; (5) validation and evaluation. To test the success of this research, a simple website-based prototype system was built, where test results were obtained using twenty reference data from the perspectives of experts in related fields, obtaining accuracy results of up to 90%. In the future, these results still need to be improved so that the solutions developed are more accurate in diagnosing pests and diseases in chili plants.