•  
  •  
 

Abstract

Early detection of shrimp diseases is vital for reducing losses and ensuring sustainable aquaculture. In this study, a novel ontology-driven framework for shrimp disease diagnosis that integrates environmental knowledge with machine learning-based image analysis is proposed. An ontology was constructed from a structured dataset of environmental parameters and farm conditions related to shrimp diseases, which formally represents concepts such as farm characteristics, environmental factors, symptoms, and disease types. Two public shrimp image datasets were subsequently used to train convolutional neural networks (ResNet50 and MobileNetV2) for automated disease classification. The predicted disease labels and confidence scores were mapped onto the ontology. Semantic reasoning rules (SWRL) and SPARQL queries were then combined with image-based predictions and environmental risk factors to generate explainable diagnostic results and tailored management recommendations. The experimental results show that this ontology–ML hybrid approach achieves higher diagnostic accuracy than traditional methods without ontology integration do, thus demonstrating its potential for providing more precise, context-aware, and actionable support for shrimp health management.

Abstract Image

Download
COinS