A Semantic-Based Framework for Rice Plant Disease Management: Identification, Early Warning, and Treatment Recommendation Using Multiple Observations

Abstract

Rice plant diseases can cause damages and yield losses. To reduce the productivity losses, farmers need to observe and decide suitable treatments for the diseases recognized from the abnormal characteristics appeared in their farms. Traditionally, farmers identify potential diseases from their experiences or by consulting other experts. However, this approach has certain disadvantages due to varying knowledge, and at times unreliable experience and perception of different farmers. Externalization of knowledge from existing reliable sources and utilization of multiple farmer’s observations can overcome such problems. Thus, this study presents the design and development of RiceMan, a semantic-based framework in agriculture for rice plant disease management using multiple observations. The framework not only manages observations within a single farm, but also integrates with neighborhood observations to cope with spreadable rice diseases. In addition, with proper design of Rice Diseases Ontology (RiceDO) and Treatment Ontology (TreatO), the framework can identify possible diseases and give early warnings to farmers for their appropriate actions. Based on realistic situations, the paper also illustrates how the proposed framework can help farmers to better: (1) identify rice diseases, (2) prepare for the early warnings, and (3) obtain recommended treatments.