Semantic Web of Things for pollution measurement and validation interoperability using AI Techniques

In response to the growing IoT device diversity, efforts are underway to better integrate data, applications, and services. The Semantic Web, known for its simplicity in integration, has the potential to improve data interpretation and interoperability. In this research, a pollution management model is used, combining the Semantic Web of Things (SWoT) and Artificial Intelligence (AI), to create smarter cities, providing real-time environmental information. The dataset has been sourced from Aarhus City, Denmark, and the study outlines Semantic Web Technologies (SWTs) in IoT frameworks, including common ontologies for IoT-based architecture. The dataset's relationship between various gases/pollutants is analyzed using correlation matrix. Machine learning methods like Multi-Layer Perceptron (MLP) with Sigmoid, ReLU, Tanh, Maxout, Swish hybrid activation functions are employed, with results assessed using Root Mean Squared Error (RMSE) and Mean Squared Error (MSE). A comparison of errors for different activation functions is also performed and the findings reveal good results when comparing actual and predicted values in the proposed model.