The Internet of Things (IoT) benefits from social networking platforms in establishing and enhancing social-oriented services, information, and autonomous social relationships. Social IoT (SIoT) systems can boost the user experience in the real world in several applications, including healthcare, transportation, and entertainment. However, the collected data from various interconnected SIoT systems is massive, demanding robust and efficient processing algorithms, feature extraction, selection, and inference. This work presents an enhanced Artificial Hummingbird algorithm (AHA) for feature selection (FS). The enhanced version of AHA is performed using the advantages of Quantum-based optimization. The main aim of using Quantum is to improve the population’s exploration ability while discovering feasible regions. Extensive experiments utilizing eighteen UCI datasets were conducted to validate the developed FS method, QAHA. The QAHA is compared with other FS methods, and the experimental established its efficiency. Moreover, a set of four datasets from SIoT are used to evaluate the applicability of QAHA to the real-world setting. The results using these datasets indicate the high performance of QAHA to increase the accuracy by decreasing the number of features. In the case of UCI datasets, the average accuracy of the developed QAHA is 93% among the eighteen datasets. Whereas, In the case of the SIoT datasets, the developed QAHA has an accuracy of nearly 90.7%, 98.7%, 92.2%, and 84.6% for the Trajectory, GAS sensors, Hepatitis, and MovementAAL datasets, respectively.