Data Pre-processing Approach with ML Algorithms for Accuracy and Authenticity Detection of Big Data Sourced from the Social Internet of Things and a Case Study

The Social Internet of Things (SIoT), which integrates sensor data with social media interactions, produces massive volumes of unstructured data that require accurate and reliable preprocessing for meaningful analysis. This study investigates the effectiveness of various machine learning (ML) classification algorithms in detecting the accuracy and authenticity of SIoT-derived data.A dataset comprising 17,500 user records collected from mobile devices and social media platforms was analyzed using five ML classifiers: Logistic Regression, Naive Bayes, K-Nearest Neighbor (K-NN), Random Forest, and Support Vector Machines (SVM). Through extensive hyperparameter tuning and 5-fold cross-validation, the Random Forest and SVM models exhibited the highest performance, achieving accuracy scores of 0.58 and 0.57, respectively. SVM also obtained the best AUC value of 0.64, highlighting its strength in distinguishing authentic from manipulated data. Additionally, the results emphasize the need for larger, more diverse datasets, and suggest incorporating deep learning and automated bias mitigation methods in future research.