Bayesian feedback-enhanced method in open set classification of ship-radiated sounds

Invasive species pose significant ecological risks, yet traditional models often neglect the inherent uncertainties arising from measurement errors, spatial aggregation, and sampling biases. This study presents a novel web-semantic framework to enhance decision-making under uncertainty by explicitly integrating a Data Quality Ontology into Maximum Entropy (MaxEnt)-based species distribution modeling workflow. The framework encompasses robust preprocessing of environmental variables such as sea temperature, salinity, and net primary production, and incorporates error propagation in feature transformations and model training. By visualizing model predictions with confidence intervals, we show how input data uncertainties affect model outputs, thereby supporting transparent risk assessment. The results demonstrate that accounting for uncertainty not only enhances the reliability of habitat suitability maps, but also provides stakeholders with actionable insights for decision making under uncertainty.