Management of network becomes increasingly difficult due to the rapid proliferation of the Internet of Things (IoT) and heterogeneous demands of applications. To mitigate congestion and maintain a high quality of service (QoS) for application users, we propose an importance-oriented clustering-based QoS system named ICAQ. The system uses unsupervised machine learning to determine and differentiate different application flow priority. The system is proposed to be a generic to both network core features and the application features. An environment monitoring IoT application is implemented as an example to demonstrate the expressive power of ICAQ. Experimental results have verified the effectiveness and efficiency of ICAQ and show a promising potential for using ICAQ for the QoS management of IoT applications in 5G and beyond networks.